Digitalization as a key driver of economic development in the music industry in European Union

54 Schriftenreihe Informatik Doucek ▪ Sonntag ▪ Nedomova (Editors) IDIMT-2025 IDIMT-2025 ICT in Business: AI Everywhere? Glory and Disgrace of AI 33rd Interdisciplinary Information Management Talks With pride we present the proceedings of the 33rd annual IDIMT Conference. Artificial Intelligence continues to make impressive advances and almost every month a new system is presented. But does this translate into a proportionate use in business – and as a second step, business advantages? In this conference we investigate various aspects of new developments to ICT itself as well as its use for management and business processes and their transformation. Ethical and security aspects as well as specific application areas (teaching, autonomous vehicles, supply chain management, social media) are touched too, to obtain an encompassing view of the topic. We have chosen the following 10 topics for 2025: ▪ Early Career & Student Showcase (Sonntag) ▪ AI in Crisis Management (Neubauer, Rainer) ▪ Cyber Security vs AI (Sonntag) ▪ Ethical Aspects of AI: Research and Usage (Lisnik) ▪ Data, AI and Digital-Driven Transformation: Shaping Sustainable Digital Futures (Pucihar) ▪ AI and Autonomous Systems (Schoitsch, Abdelkader) ▪ Business Information Systems and Digitalization (Doucek, Maryška) ▪ Social Media and AI: Contributor, Selector, …? (Pavlíček) ▪ AI in Supply Chain Management (Delina) ▪ University-Industry Collaboration (Pitner) Based on a double-blind two-step review process we have selected 43 of the submitted papers with a totality of more than 107 coauthors. The program ran in two parallel streams. The authors come from 11 different countries: Armenia, Austria, Czech Republic, Germany, Greece, Israel, Poland, Slovakia, Slovenia, Spain and Ukraine. DOI:10.35011/IDIMT-2025 Band 54 Doucek Petr ▪ Sonntag Michael ▪ Nedomova Lea (Editors) IDIMT-2025 ICT in Business: AI Everywhere? Glory and Disgrace of AI 33rd Interdisciplinary Information Management Talks Sept. 3–5, 2025 Hradec Králové, Czech Republic ISBN 978-3-99151-856-3 www.trauner.at universität VERLAG 54 Schriftenreihe Informatik Doucek Petr ▪ Sonntag Michael ▪ Nedomova Lea (Editors) IDIMT-2025 ICT in Business: AI Everywhere? Glory and Disgrace of AI 33rd Interdisciplinary Information Management Talks Sept. 3–5, 2025 Hradec Králové, Czech Republic Herausgeber: Prague University of Economics and Business nám. W. Churchilla 1938/4 130 67 Praha 3 Czech Republic Kommisionsverlag: TRAUNER Verlag + Buchservice GmbH Köglstraße 14, 4020 Linz Österreich/Austria Herstellung: paco Medienwerkstatt, 1160 Wien, Österreich/Austria 202122064 DOI: 10.35011/IDIMT-2025 ISBN 978-3-99151-856-3 www.trauner.at Impressum Schriftenreihe Informatik Doucek Petr ▪ Sonntag Michael ▪ Nedomova Lea (Editors) IDIMT-2025 ICT in Business: AI Everywhere? Glory and Disgrace of AI 33rd Interdisciplinary Information Management Talks This publication was partially supported by the Prague University of Economics, Business – project IG409035 and institutional support for the long-term conceptual development of science and research at Faculty of Informatics and Statistics (IP400040), and the Johannes Kepler University Linz. The Conference IDIMT-2025 took place September 3–5, 2025 in Hradec Králové, Czech Republic Programme Committee Abdelkader Shaaban, AT Delina Radoslav, SK Doucek Petr, CZ Lisnik Anton, SK Maryška Milos, CZ Nedomova Lea, CZ Neubauer Georg, AT Pavliček Antonín, CZ Pitner Tomáš, CZ Pucihar Andreja, SI Rainer Karin, AT Schoitsch Erwin, AT Sonntag Michael, AT Tkáč Michal, SK © 2025 The Author(s) 2025 3 TABLE OF CONTENS AI IN CRISIS MANAGEMENT POTENTIALS AND CHALLENGES OF AI IN CRISIS MANAGEMENT ACROSS AUSTRIA AND EUROPE ......................................................................................................................................13 Melissa Hagendorn, Karin Rainer, Viktoria Kundratitz, Alois Leidwein, Georg Neubauer, Dražen Ignjatović DOI: 10.35011/IDIMT-2025-13 .....................................................................................................................13 FACTORS CONTRIBUTING TO SOCIAL VULNERABILITY TO VARIOUS HAZARDS IN EUROPEAN METROPOLITAN AREAS ...........................................................................................21 Sarah Kainz, Constanze Geyer, Sofia Kirilova, Ilona Grabmaier, Benjamin Schuster, Vassiliki Apostolopoulou, Satenik Bakunts, Valeri Bagiyan, Kiril Shtefchyk, Lola Valles, Ioanna Triantafyllou, Danai Kazantzidou, John Tsaloukidis DOI: 10.35011/IDIMT-2025-21 .....................................................................................................................21 MATCHING NEEDS AND CAPABILITIES FOR PANDEMIC MANAGEMENT .........................29 Georg Neubauer, Dražen Ignjatović, Verena Parzer, Karin Rainer, Viktoria Kundratitz, Melissa Hagendorn, Georg Aumayr, Sabine Kretschy DOI: 10.35011/IDIMT-2025-29 .....................................................................................................................29 THE USE OF AN AI-SUPPORTED TOOL FOR THE DEPLOYMENT OF INFORMAL VOLUNTEERS IN CRISIS AND DISASTER MANAGEMENT ON THE BASIS OF THEIR COMPETENCIES ................................................................................................................................37 Sabine Kretschy, Nadine Sturm, Martin Söllner, Christoph Angster, Andreas Rath, Bernhard Bürger, David Schneeberger, Georg Hahn, Birgit Pröll, Sandra Pichler DOI: 10.35011/IDIMT-2025-37 .....................................................................................................................38 AI IN CRISIS MANAGEMENT: BIBLIOMETRIC ANALYSIS ......................................................45 Petr Řehoř, Lukáš Klarner DOI: 10.35011/IDIMT-2025-45 .....................................................................................................................45 CYBER SECURITY VS AI A SECURE SCHEME FOR CHAINED AUTHENTICATION COMBINED WITH ATTESTATION ...................................................................................................................................55 Michael Sonntag DOI: 10.35011/IDIMT-2025-55 .....................................................................................................................55 AI, CYBERSECURITY AND THE REAL VALUE PROPOSITION FOR BUSINESS: MOVING BEYOND THE HYPE ..........................................................................................................................63 Martin Zbořil DOI: 10.35011/IDIMT-2025-63 .....................................................................................................................63 APPLYING PROTECTION MOTIVATION THEORY TO ENHANCE ADHERENCE TO ONLINE SECURITY PRACTICES ....................................................................................................71 Tomáš Sigmund DOI: 10.35011/IDIMT-2025-71 .....................................................................................................................71 4 ETHICAL ASPECTS OF AI: RESEARCH AND USAGE SELECTED ETHICAL ASPECTS OF AUTHORSHIP OF THE USE OF AI PRODUCTS ...........81 Anton Lisnik DOI: 10.35011/IDIMT-2025-81 .....................................................................................................................81 PERCEPTIONS AND ATTITUDES OF UNIVERSITY STUDENTS TOWARDS ETHICAL ASPECTS OF ARTIFICIAL INTELLIGENCE .................................................................................89 Ivan Katrenčík, Martina Kuperová DOI: 10.35011/IDIMT-2025-89 .....................................................................................................................89 ETHICAL ASPECTS OF ARTIFICIAL INTELLIGENCE AND THE DEVELOPMENT OF ENTREPRENEURSHIP AMONG YOUNG PEOPLE .......................................................................97 Krzysztof M. Krusiec DOI: 10.35011/IDIMT-2025-97 .....................................................................................................................97 AI AND CVS – AN ANALYSIS OF THE DISCREPANCIES BETWEEN GENERATED PROFILES AND CANDIDATES' ACTUAL SKILLS ...................................................................... 105 Andrzej Palej DOI: 10.35011/IDIMT-2025-105 ................................................................................................................. 105 VIRAL MARKETING BEHAVIOUR OF GENERATION Z .......................................................... 113 Martina Kuperová, Monika Zatrochová DOI: 10.35011/IDIMT-2025-113 ................................................................................................................. 113 RESPONSIBLE MARKETING IN TOURISM: USING AI WITHOUT ETHICAL VIOLATIONS .................................................................................................................. 121 Patrik Bretz, Miroslav Warhol DOI: 10.35011/IDIMT-2025-121 ................................................................................................................. 121 DATA, AI AND DIGITAL-DRIVEN TRANSFORMATION: SHAPING SUSTAINABLE DIGITAL FUTURES OPEN DATA READINESS: EXPLORATORY STUDY OF SLOVENIAN ENTERPRISES ........ 131 Mirjana Kljajić Borštnar, Andreja Pucihar DOI: 10.35011/IDIMT-2025-131 ................................................................................................................. 131 UTILIZATION OF AI-DRIVEN MANUFACTURING TECHNOLOGIES: AN EMPIRICAL STUDY .................................................................................................................. 139 Ján Závadský, Zuzana Závadská DOI: 10.35011/IDIMT-2025-139 ................................................................................................................. 139 DIGITALIZATION AS A KEY DRIVER OF ECONOMIC DEVELOPMENT IN THE MUSIC INDUSTRY IN EUROPEAN UNION ................................................................................................ 147 Maria Bartekova, Sabina Janikovicova DOI: 10.35011/IDIMT-2025-147 ................................................................................................................. 147 MODERN DATA ARCHITECTURES: EVALUATION AND SELECTION CRITERIA FOR DATA-DRIVEN ENTERPRISES ...................................................................................................... 155 Felix Espinoza, Milos Maryska DOI: 10.35011/IDIMT-2025-155 ................................................................................................................. 155 5 AI-ASSISTED DATA GOVERNANCE: FROM THEORY TO TOOLS ........................................ 163 Stepan Stanek DOI: 10.35011/IDIMT-2025-163 ................................................................................................................. 163 ECOSPELL: AN ECO-FRIENDLY PREDICTIVE MAINTENANCE FRAMEWORK FOR THE INDUSTRIAL INTERNET OF THINGS (IIOT) .............................................................................. 173 Nikola Kuchtíková, Milos Maryska DOI: 10.35011/IDIMT-2025-173 ................................................................................................................. 173 AI AND AUTONOMOUS SYSTEMS HUMAN-MACHINE TEAMING – THE WAY FORWARD TO A SUPER-SMART SOCIETY .. 183 Erwin Schoitsch DOI: 10.35011/IDIMT-2025-183 ................................................................................................................. 183 CYBERSECURITY COMPETENCIES FOR THE FUTURE – INSIGHTS FROM THE PROJECT CYBERSECURITY-KOMPETENZ AUSTRIA (CSKA) ...................................... 193 Christoph Schmittner, Abdelkader Shaaban, Emanuel Tananau Blumenschein, Roger Von Laufenberg, Stefan Hopf, Hannes Dangl, Stephan Blahut DOI: 10.35011/IDIMT-2025-193 ................................................................................................................. 193 BUSINESS INFORMATION SYSTEMS AND DIGITALIZATION DIGTALIZATION IN BUSINESS ..................................................................................................... 205 Petr Doucek, Lea Nedomova, Milos Maryska DOI: 10.35011/IDIMT-2025-205 ................................................................................................................. 205 PERSPECTIVE ON THE ISSUE OF OPEN INNOVATION IN THE ERA OF ACCELERATED DIGITALIZATION AND ARTIFICIAL INTELLIGENCE ............................................................ 215 Zuzana Dzilská, František Pollák, Michal Konečný DOI: 10.35011/IDIMT-2025-215 ................................................................................................................. 215 SUPPORT OF DIGITAL TRANSFORMATION OF HUMAN RESOURCE PROCESSES AT CITY HALL BY ENTERPRISE ARCHITECTURE APPROACH: A CASE OF THE REGIONAL CZECHIA ADMINISTRATION ......................................................................... 223 Martin Lukáš, Miroslav Brabec DOI: 10.35011/IDIMT-2025-223 ................................................................................................................. 223 CURRENT DEVELOPMENTS IN THE INDUSTRY FROM THE PERSPECTIVE OF THE RUBBER PROCESSING SECTOR ................................................................................................... 233 Miroslav Dusík DOI: 10.35011/IDIMT-2025-233 ................................................................................................................. 233 FROM OVERLOAD TO ARTIFICIAL INTELLIGENCE. MAPPING THE DETERMINANTS OF TECHNOSTRESS IN MODERN WORK ENVIRONMENTS .................................................. 247 Martina Rašticová, Nataliia Tkalenko, Filip Brutovský, Petia Genkova, Sylwia Przytula, Nataliia Versal DOI: 10.35011/IDIMT-2025-47 ................................................................................................................... 247 6 SOCIAL MEDIA AND AI: CONTRIBUTOR, SELECTOR, …? AI INNOVATIONS - THE DOUBLE-EDGED SWORD: EXPLORING POSITIVE AND NEGATIVE IMPLICATIONS OF ARTIFICIAL INTELLIGENCE IN SOCIAL MEDIA ........... 257 Antonín Pavlíček DOI: 10.35011/IDIMT-2025-257 ................................................................................................................. 257 INVESTIGATING GAMIFICATION AS A SOLUTION FOR READING FATIGUE .................. 269 Majid Ziaei Nafchi DOI: 10.35011/IDIMT-2025-269 ................................................................................................................. 269 GAI-LENS: LEGAL AND ETHICAL RISK ASSESSMENT IN GENERATIVE AI SOCIAL CONTENT USING ADVANCED AI METHODS ............................................................................. 277 Hanan Maoz DOI: 10.35011/IDIMT-2025-277 ................................................................................................................. 277 RISK AVOIDANCE IN DIGITAL INTERACTION: RISK-AVERSION AND DEMOGRAPHIC DETERMINANTS OF SOCIAL NETWORK USAGE..................................................................... 285 František Sudzina, Antonín Pavlíček DOI: 10.35011/IDIMT-2025-285 ................................................................................................................. 285 MISINFORMATION AND INFLUENCE OPERATIONS IN ELECTIONS: GLOBAL TRENDS AND THE 2024 ROMANIAN CASE ................................................................................................. 293 Jiří Korčák, Richard A. Novák, David Pavlů DOI: 10.35011/IDIMT-2025-293 ................................................................................................................. 293 MORE IS LESS? EXPLORING THE DILUTION EFFECT OF GROWING PUBLICATION NETWORK ON CITATION RATES IN HARD SCIENCES........................................................... 301 Lubomír Štěpánek DOI: 10.35011/IDIMT-2025-301 ................................................................................................................. 301 FACEBOOK POSTS VIRALITY - SHARING OR LIKING? ......................................................... 309 Jana Syrovátková DOI: 10.35011/IDIMT-2025-309 ................................................................................................................. 309 SOCIAL MEDIA AND THE EVOLUTION OF CROWDSOURCING: FROM USERGENERATED CONTENT TO AI-ASSISTED CREATIVITY ........................................................ 317 Jan Lešetický, Lukáš Malec DOI: 10.35011/IDIMT-2025-317 ................................................................................................................. 317 AI IN SUPPLY CHAIN MANAGEMENT THE POTENTIAL OF SOCIAL ENTERPRISES IN SLOVAK PUBLIC PROCUREMENT: DESCRIPTIVE ANALYSIS AND AI-DRIVEN MATCHING FRAMEWORK .............................. 327 Radoslav Delina, Michaela Šulová DOI: 10.35011/IDIMT-2025-327 ................................................................................................................. 327 CLUSTERING GLOBAL INNOVATION ACTIVITY IN LARGE LANGUAGE MODELS VIA PATENT DATA .......................................................................................................................... 337 Martin Potančok, Jan Černý DOI: 10.35011/IDIMT-2025-337 ................................................................................................................. 337 STRUCTURED INSIGHTS INTO GREEN PROCUREMENT: THE LITHUANIAN CASE........ 345 Mariana Ivaničková, Miroslava Barkóciová, Michal Tkáč DOI: 10.35011/IDIMT-2025-345 ................................................................................................................. 345 7 AN EMPIRICAL SURVEY OF SKILLS AND COMPETENCY EXPECTATIONS AMONG STUDENTS ......................................................................................................................................... 353 Markéta Zajarošová, Jakub Karas, Radoslav Delina DOI: 10.35011/IDIMT-2025-353 ................................................................................................................. 353 AI-SUPPORTED DECISION-MAKING IN THE OUTSOURCING OF INTERNAL LOGISTICS PROCESSES ................................................................................................................. 361 Erik Weiss, Mária Janošková, Roland Weiss DOI: 10.35011/IDIMT-2025-361 ................................................................................................................. 361 UNIVERSITY-INDUSTRY COLLABORATION ROLE OF ACADEMIA IN SEMICONDUCTOR INNOVATIONS: BETWEEN FREEDOM AND RESILIENCE ............................................................................................................................ 373 Jan Ministr, Tomáš Pitner DOI: 10.35011/IDIMT-2025-373 ................................................................................................................. 373 INTEGRATION OF EDUCATION AIMED AT DEVELOPING THE KNOWLEDGE AND SKILLS OF THE STUDENT REFLECTING THE REQUIREMENTS OF THE 21ST CENTURY. ........................................................................................................................ 381 Zuzana Chodasová, Miriam Takáčová DOI: 10.35011/IDIMT-2025-381 ................................................................................................................. 381 LABOUR MARKET EVOLUTION IN LIGHT OF DEMOGRAPHIC CHANGE ......................... 391 Jaromír Veber, Marek Botek DOI: 10.35011/IDIMT-2025-391 ................................................................................................................. 391 EARLY CAREER & STUDENT SHOWCASE GENDER EQUALITY IN HIGH-TECH SECTORS: ANALYSIS OF WOMEN’S REPRESENTATION IN TECHNOLOGY INDUSTRIES IN THE VISEGRÁD GROUP COUNTRIES ...................................................................................................................................... 401 Sabina Janikovicova, Diana Pallérová, František Pollák, Petra Pártlová DOI: 10.35011/IDIMT-2025-401 ................................................................................................................. 401 PUBLICLY AVAILABLE IT KPIS FOR BUSINESS PERFORMANCE EVALUATION ............ 409 Adéla Zbořilová DOI: 10.35011/IDIMT-2025-409 ................................................................................................................. 409 USING ARCHIMATE TO ALIGN AGILE SOFTWARE PRODUCT MANAGEMENT WITH ENTERPRISE ARCHITECTURE .................................................................................................... 417 Adam Krbusek, Alena Buchalcevova DOI: 10.35011/IDIMT-2025-417 ................................................................................................................. 417 ANNEX STATEMENT OF THE PUBLICATION ETHICS AND PUBLICATION MALPRACTICE........ 429 LIST OF AUTHORS .......................................................................................................................... 431 8 15 The project highlighted critical challenges in AI application. Although testing was not conducted in live emergencies, realistic scenarios exposed limitations in training data, operational readiness, and speech recognition reliability, particularly under conditions of dialect, urgency, and emotional stress. Legal and ethical obligations, including GDPR compliance and AI Act transparency, had to be addressed from the outset. Trust in the system depended on its explainability, fallback mechanisms, and operational resilience under uncertain conditions. Future work will focus on enhancing explainability, multimodal robustness, and interoperability with emergency systems, aiming to develop adaptive AI agents grounded in trust-by-design principles and suited to high-stakes communication environments. 2.2 Evidence-based planning and AI potential for pandemic management: the ROADS project While KRISAN focused on AI-supported communication in crisis management, the ROADS project (https://www.ages.at/en/roads) developed a demonstrator concept to support pandemic planning by aligning public health objectives with corresponding social measures. Although AI was not yet applied, ROADS laid essential groundwork for future integration. The Austrian Ministry of Health and relevant stakeholders expressed interest in advancing the ROADS system into an AI-supported solution, recognising its potential to manage complex, high-volume data and strengthen evidence-based decision-making. The project also underscored the necessity of expert validation for any AI-supported outputs (Rainer et al., 2024). Based on legal, strategic, and policy analyses, ROADS introduced an expert-driven matching process to link strategic objectives to interventions, exposing challenges such as unclear strategic mandates, fragmented data sources, and limited institutional coordination (Rainer et al., 2023). Two follow-up directions are under consideration: one targets adaptive AI integration for dynamic alignment of policy goals and real-time data, building on automated knowledge synthesis and crossinstitutional data harmonisation (Fontes et al., 2023). The other addresses usability and stakeholder alignment, aiming to strengthen trust and legitimacy through co-design, interpretability, and iterative feedback loops (Rainer et al., 2023). Both directions prioritise expert validation and contextual sensitivity, ensuring that AI augments, rather than replaces, human judgement in future crisis response systems (Shneiderman, 2020). 3. Challenges in AI application across Austria and Europe National projects such as KRISAN and ROADS illustrate recurring challenges in applying AI to crisis management. These include fragmented data environments, siloed research funding, and limited alignment between research capacities and regulatory demands in Austria. Similar difficulties persist across Europe, where short-term project cycles and low institutional readiness continue to obstruct long-term AI adoption (Gao & Janssen, 2022). Regulatory frameworks within the European Union, especially the AI Act, are increasingly seen as a global benchmark for responsible AI governance (Stix, 2021). However, a growing gap remains between regulatory ambition and the practical implementation of research outcomes, particularly within public infrastructure. While research activity is accelerating globally, the institutional conditions required to translate AI innovation into sustainable public-sector applications are still maturing across much of Europe. 16 3.1 Common challenges at the national and European level Strengthening alignment between institutional capacity, governance, and innovation strategies will be crucial for ensuring that AI systems contribute effectively and responsibly to crisis management. Table 1 outlines core implementation challenges from KRISAN and ROADS, alongside recurring patterns across the European AI landscape. Source: (authors) 3.2 Opportunities and risks of AI in crisis-related applications The challenges summarised in Table 1, along with growing awareness among AI users of the need to address them proactively and holistically, form the basis for unlocking AI’s potential in health-related crisis management. When such limitations are actively mitigated, artificial intelligence can support more responsible and context-sensitive applications. Conversely, overlooking these issues, or lacking the literacy to recognise them, may amplify existing risks. Table 1. Clustered challenges in AI implementation across Austria and Europe challenge type Austria (projects KRISAN and ROADS) Europe (literature review) technical low explainability and interpretability limited reliability due to data instability limited speech recognition accuracy fragmented innovation strategies lack of scalable, robust solutions (Serger et al., 2023) legal and regulatory real-time GDPR and AI Act compliance (see also GDPR, 2016; AI Act, 2024) unclear legal responsibilities in emergency use (Hagendorn et al., 2024) gap between high-level AI principles and concrete, implementable policy actions misalignment in health and technological governance structures (Van Noordt & Medaglia, 2023) ethical and organisational gaps in transparency, trust, and accountability widen global inequalities lack of consistent organisational guidelines (BMKOES, 2023) inconsistent integration of ethics presence of gender bias difficulty in translating ethical principles into actionable guidelines (Bird et al., 2020; González & Rampino, 2024; Kiseleva et al., 2022) siloed sectors prevent coordinated AI development (Starke et al., 2025) systemic and environmental short-term project funding vs. long-term application and adaptation needs lack of AI policy coherence unaddressed environmental impacts (Rainer et al., 2023) structural funding constraints similar to Austrian issues intensive energy-use and high carbon emissions (Strubell et al., 2019) 17 Figure 1 presents a simplified overview of these opportunities and risks, drawing on insights from academic and institutional sources (Kelly et al., 2022; EPRS, 2023; WHO, 2023). Figure 1. Core opportunities and risks of AI applicable to crisis management and decision support Source: (authors, figure generated using napkin.ai) These factors are relevant not only for technological development but also for regulatory coherence and institutional adoption across Europe. Opportunities such as increased operational efficiency, targeted automation, and system-level responsiveness have the potential to enhance decision-making and resource coordination in dynamic scenarios (Vinuesa et al., 2020). While AI can accelerate insights and support timely action, it also raises concerns about overreliance on automated outputs, reduced critical oversight, and the risk of human deference to system recommendations, particularly under time pressure or in emotionally charged situations (Leslie, 2019). Uneven implementation of data protection continues to affect public trust. Similarly, the impact of AI on labour markets remains contested, with debates ranging from role transformation to potential displacement. 4. Understanding and strengthening AI literacy As AI systems become increasingly integrated into public life, AI literacy has emerged as a foundational competence for individuals and institutions navigating an increasingly automated world. It encompasses the ability to understand, evaluate, and interact meaningfully with AI systems. As defined by Long and Magerko (2020), this includes interpreting how algorithms function, assessing their outputs, and engaging critically with their limitations and potential biases. Building on this, Kumar and Sangwan (2024) conceptualise AI literacy as a prerequisite for responsible AI use, particularly in domains where trust, accountability, and societal impact are at stake. In high-stakes environments such as crisis management and public health, AI literacy enables informed oversight and mitigates overreliance on opaque “black box” systems whose outputs might otherwise be accepted without question. Yet a persistent competence gap is evident among practitioners, including emergency responders and healthcare professionals, who often lack the training required to interpret or supervise AI tools effectively (Hassan et al., 2024). This unmet need extends beyond institutional actors. As noted by Sun and Medaglia (2019) and Wolniak and Stecula (2024), limited AI awareness also affects the general public, including those directly impacted by crises, whose capacity to understand or challenge automated decisions remains 18 minimal. This lack of comprehension compounds existing digital divides, especially in vulnerable population groups (Saka, Hormiga, & Valls-Pasola, 2024). An important aspect of AI literacy is transparency in human-AI interaction. In public-facing systems such as crisis helplines or digital health triage systems, it is essential that users are explicitly informed when they are interacting with an AI rather than a human agent. Disclosure protocols and opt-out mechanisms help preserve user agency and trust. The consequences of low AI literacy extend beyond individual misunderstanding. It can weaken public trust (Aoki, 2020), reduce institutional transparency, and undermine ethical governance frameworks (Kuziemski & Misuraca, 2020). As Kholov and Mamarasulov (2025) argue, long-term AI deployment in public administration depends on sustained awareness and competence. Educational and embedded training programmes are essential to closing these competency gaps (Achanta, 2025). Beyond formal training, everyday practices can help foster AI literacy among both professionals and the general public. One such practice is AI prompting – the act of interacting with systems through tailored inputs and queries. Prompting encourages exploration, facilitates gradual competence development, and fosters critical engagement with AI tools. While prompting was not a central concept in KRISAN, its structured dialogue flows illustrate how interaction design can foster user agency and foundational prompting skills, such as phrasing questions clearly, interpreting responses, and managing expectations. Moreover, normalising such interaction helps destigmatise AI use, reframing it not as a threat or novelty but as a tool that, when well understood, can empower individuals and support decision-making. Ultimately, strengthening AI literacy is not an optional addon but a structural prerequisite for responsible, inclusive, and effective AI use in crisis management and beyond. 5. Conclusion and outlook Despite promising advances in research and demonstrators, the broader-scale application of artificial intelligence in crisis management, particularly in public health, continues to face significant challenges. Technical opacity, fragmented data ecosystems, legal uncertainties, and low institutional readiness impair the responsible deployment of AI in high-stakes domains such as public health. These concerns underscore the need for more mature, adaptive, and transparent AI systems. At the same time, AI offers considerable potential to transform decision-making through faster data analysis, predictive modelling, and dynamic coordination. While KRISAN demonstrates applied use cases, ROADS provides strategic foundations for understanding how AI could be embedded into pandemic management. Together, they underscore the complexity of aligning AI with operational and strategic demands. Addressing these systemic and technical challenges also requires investment in human capabilities. Responsible innovation in AI depends on broad-based literacy that enables high-level decision-makers, practitioners, and citizens to understand, question, and apply AI systems effectively. Strengthening AI literacy across disciplines and institutions is vital. Structured prompting and handson engagement can provide accessible opportunities to build this capacity. Future efforts should prioritise interdisciplinary collaboration, long-term investment strategies, and cross-border coordination to address shared challenges. With deliberate, context-sensitive governance and inclusive engagement, AI can become a powerful enabler rather than a risk factor in building resilient and equitable crisis response systems. Ultimately, the glory or disgrace of AI in crisis management is not intrinsic to the technology itself, but a reflection of the choices we make in how we develop, regulate, and apply these technologies in service of the public good. 19 Acknowledgement The research leading to these results has received funding from KIRAS Cooperative R&D Projects 2021 KRISAN with project number FO99901586 and ROADS with project number FO999899442. References Achanta, A. (2025). AI in Public Services: Pathways to Literacy and Trust. In AI in Public Services. IGI Global. DOI: 10.4018/979-8-3693-8372-8.ch005 AI Act: Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024. https://eurlex.europa.eu/eli/reg/2024/1689/oj, retrieved 22.04.2025 Aoki, N. (2020). An experimental study of public trust in AI chatbots in the public sector. Government Information Quarterly, 37(4), 101509. https://doi.org/10.1016/j.giq.2020.101509 Bird, E., Fox-Skelly, J., Jenner, N., Larbey, R., Weitkamp, E., Winfield, A., Panel for the Future of Science and Technology, European Parliamentary Research Service & Scientific Foresight Unit (STOA). (2020). The ethics of artificial intelligence: Issues and initiatives. In STOA | Panel For The Future Of Science And Technology (report PE 634.452). https://doi.org/10.2861/6644 Bundesministerium für Kunst, Kultur, öffentlichen Dienst und Sport (BMKOES). (2023). Leitfaden Digitale Verwaltung und Ethik: Praxisleitfaden für KI in der Verwaltung (Version 1.0, pp. 32-34). https://oeffentlicherdienst.gv.at/wp-content/uploads/2023/11/Leitfaden-Digitale-Verwaltung-Ethik.pdf, retrieved 22.04.2025 EPRS – European Parliamentary Research Service. (2023). Understanding the EU’s AI Act: Scope, approach and implications. https://www.europarl.europa.eu/thinktank/en/document/EPRS_BRI(2023)739123, retrieved 22.04.2024 Fontes, A. C., Corrigan, C. C. & Lütge, C. (2023). Governing AI during a pandemic crisis: Initiatives at the EU level. Technology in Society, 72, 102204. https://doi.org/10.1016/j.techsoc.2023.102204 Gao, Y., & Janssen, M. (2022). The Open Data Canvas: Analyzing value creation from open data. Digital Government: Research and Practice, 3(1), Article 5, 1–15. https://doi.org/10.1145/3511102 General Data Protection Regulation (GDPR). (2016). Regulation (EU) 2016/679. https://gdpr-info.eu/, retrieved 22.04.2025 González, A. S., Rampino, L. (2024). A design perspective on how to tackle gender biases when developing AI-driven systems. AI Ethics. https://doi.org/10.1007/s43681-023-00386-2 Hagendorn, M., Rainer, K., Kundratitz, V., Leidwein, A., Fuhrmann, F. (2024) AI-based voice dialogue system for enhancing crisis management in the KRISAN project. In P. Doucek, M. Sonntag, & L. Nedomová (Eds.), IDIMT2024: Information Technology, Society and Economy – Strategic Cross-Influences. 32nd interdisciplinary Information Management Talks. (pp. 163-174). Trauner Verlag. https://doi.org/10.35011/IDIMT-2024-43 Hassan, S., et al. (2024). Health Professionals and Trust in AI: A Human Factors Perspective. JMIR Human Factors, 2024(1), e48633. doi: 10.2196/48633 Kelly, S., Kaye, S. A., Oviedo-Trespalacios, O. (2022). What factors contribute to the acceptance of artificial intelligence? A systematic review. Telematics and Informatics, Volume 77, 2023, 101925. https://doi.org/10.1016/j.tele.2022.101925 Kholov, A., & Mamarasulov, S. (2025). Problems of Using Artificial Intelligence and Digital Transformation in the System of Public Administration. Artsakhlib.am. https://artsakhlib.am/wp-content/uploads/2025/02/A.-KholovS.-Mamarasulov-Problems-of-Using-Artificial-Intelligence-and-Digital-Transformation-in-the-System-of-PublicAdministration.pdf Kiseleva, A., Kotzinos, D., De Hert, P. (2022). Transparency of AI in Healthcare as a Multilayered System of Accountabilities: Between Legal Requirements and Technical Limitations. Frontiers in Artificial Intelligence, Volume 5. https://doi.org/10.3389/frai.2022.879603 Kumar, A., & Sangwan, N. (2024). Conceptualising AI literacy: Educational and policy initiatives for a future-ready society. Goldsmiths Research Online. https://research.gold.ac.uk/id/eprint/35941/ 20 Kuziemski, M., & Misuraca, G. (2020). AI governance in the public sector: Three tales from the frontiers of automated decision-making in democratic settings. Telecommunications Policy, 44(6), 101976. https://doi.org/10.1016/j.telpol.2020.101976 Leslie, D. (2019). Understanding artificial intelligence ethics and safety. The Alan Turing Institute. https://www.turing.ac.uk/sites/default/files/2019-06/understanding_artificial_intelligence_ethics_and_safety.pdf Long, D., & Magerko, B. (2020). What is AI literacy? Competencies and design considerations. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems (pp. 1–16). Association for Computing Machinery. https://doi.org/10.1145/3313831.3376727 Rainer, K., Grosser, C., Hagendorn, M., Hafner, A., Leidwein, A., Neubauer, G., Sturm, N., Geyer, C., Salomon, G., Prüller, R., & Schöggl, F. (2023). From strategic planning to tackling pandemics. In P. Doucek, M. Sonntag, & L. Nedomova (Eds.), IDIMT-2023: New Challenges for ICT and Management (Vol. 52, pp. 29–38). Linz: Trauner Verlag. https://doi.org/10.35011/IDIMT-2023 Rainer, K., Kundratitz, V., Hagendorn, M., Neubauer, G., Ignjatović, D., Kutalek, R., & Sturm, N. (2024). Artificial intelligence in crisis management: Socio-technical potentials and critical uncertainties. In P. Doucek, M. M. Cehlar, & M. Basl (Eds.), IDIMT-2023: Strategic modeling in a digital environment. 31st Interdisciplinary Information Management Talks (pp. 15–27). Trauner Verlag. https://doi.org/10.35011/idimt-2023-02 Saka, T. N., Hormiga, E., & Valls-Pasola, J. (2024). Crisis response strategies: A digital reluctance perspective. Review of Managerial Science. Advance online publication. https://doi.org/10.1007/s11846-024-00822-5 Serger, S. S., Eklund, J., Remøe, S., & Åström, T. (2023). Making AI work for Sweden: Policy, research, and innovation strategies for an AI-powered future. Vinnova. Shneiderman, B. (2020). Human-Centered Artificial Intelligence: Reliable, safe & Trustworthy. In arXiv.org. https://arxiv.org/ftp/arxiv/papers/2002/2002.04087.pdf Starke, G., Gille, F., Termine, A., Aquino, Y. S. J., Chavarriaga, R., Ferrario, A., Hastings, J., Jongsma, K., Kellmeyer, P., Kulynych, B., Postan, E., Racine, E., Sahin, D., Tomaszewska, P., Vold, K., Webb, J., Facchini, A., & Ienca, M. (2025). Finding consensus on trust in AI in health care: Recommendations from a panel of international experts. Journal of Medical Internet Research, 27, e56306. https://doi.org/10.2196/56306 Stix, C. (2021). Actionable principles for artificial intelligence policy: Three pathways. Nature Machine Intelligence, 3(12), 1040–1043. https://doi.org/10.1038/s42256-021-00419-9 Strubell, E., Ganesh, A. & McCallum, A. (2019). Energy and Policy Considerations for Deep Learning in NLP (05.06.2019). arXiv.org. https://arxiv.org/abs/1906.02243 Sun, T. Q., & Medaglia, R. (2019). Mapping the challenges of Artificial Intelligence in the public sector: Evidence from public healthcare. Government Information Quarterly, 36(2), 368–383. https://doi.org/10.1016/j.giq.2018.09.008 Van Noordt, C., Medaglia, R., & Tangi, L. (2023). Policy initiatives for artificial intelligence-enabled government: An analysis of national strategies in Europe. Public Policy and Administration. https://doi.org/10.1177/09520767231198411 Vinuesa, R., Azizpour, H., Leite, I., Balaam, M., Dignum, V., Domisch, S., Felländer, A., Langhans, S. D., Tegmark, M., & Fuso Nerini, F. (2020). The role of artificial intelligence in achieving the Sustainable Development Goals. Nature Communications, 11, 233. https://doi.org/10.1038/s41467-019-14108-y Wolniak, R., & Stecuła, K. (2024). Understanding artificial intelligence in healthcare: User literacy and implications. AI, 7(3), 57. https://www.mdpi.com/2624-6511/7/3/57 World Health Organization. (2023). Regulatory considerations on artificial intelligence for health. WHO. https://www.who.int/publications/i/item/9789240078871, retrieved 22.04.2025 21 FACTORS CONTRIBUTING TO SOCIAL VULNERABILITY TO VARIOUS HAZARDS IN EUROPEAN METROPOLITAN AREAS Sarah Kainz, Constanze Geyer, Sofia Kirilova, Ilona Grabmaier, Benjamin Schuster Johanniter Österreich Ausbildung und Forschung gemeinnützige GmbH [email protected]; Constanze.Gey[email protected]om; [email protected]; [email protected]; [email protected] Vassiliki Apostolopoulou PRACTIN [email protected] Satenik Bakunts, Valeri Bagiyan Scientific Research Center, Crisis Management State Academy [email protected]; [email protected] Kiril Shtefchyk, Lola Valles Institute for Public Security of Catalonia [email protected]; [email protected] Ioanna Triantafyllou Institute of Physics of the Earth’s Interior & Geohazards, Hellenic Mediterranean University [email protected] Danai Kazantzidou, John Tsaloukidis Center for Security Studies (KE.ME.A.) [email protected]; [email protected] DOI: 10.35011/IDIMT-2025-21 22 Keywords Vulnerability; social vulnerability; indicators; vulnerable groups; hazards; disaster risk management Abstract Disasters tend to disproportionately impact individuals and communities with pre-existing social vulnerabilities. Assessing social vulnerability of areas, communities, or individuals, is not only essential for providing effective protection and care during disasters, but also for reducing vulnerabilities and enhancing resilience. In the context of the EU-funded project PANTHEON, a study was conducted which resulted in a list of possible social vulnerability indicators to assess the vulnerability of a community or of individuals to various hazards. These factors were specifically tailored to the European metropolises Paris, Athens, and Vienna. This paper will present the identified indicators and discuss their benefits for disaster risk management (DRM). Besides an extensive literature review, the development of the list was informed by expert interviews and expert surveys. The research resulted in a non-exhaustive list of 16 social vulnerability factors, covering life-stage-, health-, social-connection-, resource-, exposure-and-protection-, and knowledge-andawareness-related vulnerabilities. This list of social vulnerability indicators could be used by researchers, disaster risk managers, and other stakeholders to evaluate social vulnerability at both individual and community levels to support targeted risk reduction and resilience-building efforts, potentially also using artificial intelligence (AI). 1. Introduction This publication was adapted from the publicly available Deliverable D2.3 (Kainz et al., 2024) of the EU-funded project PANTHEON. The indicators were adapted and revised for the scope of this publication, which focuses on social vulnerability factors to disasters. PANTHEON aims to develop a community-based Smart City Digital Twin Platform for optimised DRM operations and enhanced community disaster resilience. Further information on the project can be found on the website (https://pantheon-project.eu/). 1.1 Vulnerability to hazards in the European context Vulnerability is one of the key concepts in disaster research. It is defined by the United Nations as “conditions determined by physical, social, economic and environmental factors or processes which increase the susceptibility of an individual, a community, assets or systems to the impacts of hazards” (United Nations General Assembly, 2016, p. 24). When disasters occur, the extent of damage, casualties, and recovery capacity is heavily influenced by these factors (Choo & Yoon, 2024; Cutter, 1996). While there are many aspects that increase vulnerability, like physical, economic, or environmental factors, this study focuses on social vulnerability factors that make certain individuals and communities more susceptible to disaster impacts. Disasters disproportionally affect those who are already in vulnerable situations, while also exacerbating existing vulnerabilities by disrupting livelihoods, healthcare access, or social support networks. In the past, DRM strategies have often overlooked the needs of vulnerable groups, but awareness of this issue has increased in recent years. The Council of Europe, for instance, emphasizes the need to address social vulnerabilities as part of DRM (Prieur, 2012). Several social groups are frequently listed in the literature as particularly vulnerable to disasters, including people with 23 disabilities, migrants, asylum seekers and refugees, children, the elderly, people living in poverty, and women (Centers for Disease Control and Prevention, 2022; Council of Europe, n.d.; Dearden, 2017; Flanagan et al., 2011; Painter et al., 2024). However, vulnerability is context-dependent; for instance, while women are often at greater risk in hazard situations, there are cases where men may be more vulnerable, e.g. flooding (due to increased risk-taking behaviour) or road accidents (European Commission, 2022; Jonkman & Kelman, 2005). 1.2 Vulnerability indicators Indicators are functions that map observable variables, such as the number of elderly residents in an area, to broader theoretical concepts, such as vulnerability to disasters. Generally, indicators are simple linear functions with monotonous increase or decrease. Since vulnerability is a theoretical concept that cannot be measured directly, vulnerability indicators make the concept operational instead of measuring it (Hinkel, 2011). Vulnerability indicators can be used to identify particularly vulnerable people, regions or sectors on a local scale (summarized in Hinkel, 2011). In the literature, different indices for measuring the social vulnerability to hazards can be found. For instance, Jublee and Saikat Kumar (2016) used principal component analysis to identify socioeconomic and infrastructure-related vulnerabilities in India, including the degree of homelessness and marginalization and the access to basic services. Similarly, Scheuer et al. (2011), who examined vulnerability to flooding in Leipzig, Germany, divided vulnerability into economic, social, and ecological factors. These categories included elements such as the number of children and elderly people, social and health care related infrastructure, and unemployment rate, in a particular area. Streifeneder et al. (2024) on the other hand, focusing on the COVID-19 pandemic in Austria, identified four categories: Biological susceptibility (health condition), generic susceptibility (including factors like age and living conditions), lack of capacity to anticipate (level of formal education), and lack of capacity to cope (including e.g. their access to health facilities). Building on a combination of prior existing frameworks and expert opinions, this study aims to develop a list of social vulnerability indicators applicable to European metropolitan areas, based largely on data from Athens, Paris, and Vienna. These areas were chosen since they are the focus areas of the project PANTHEON. 2. Material and Methods The indicators were derived through a methodology comprised of expert interviews, expert surveys, and an extensive literature review. To comply with the focus regions within the project PANTHEON, we focused on the areas of Athens, Paris, and Vienna, with some participants being from other areas in Greece, France, and Austria. Participants were recruited from two groups, namely DRM Stakeholders (Civil Protection Authorities, First Responders and emergency services, utilities and infrastructure providers, private companies, media, donors, governmental and policy making authorities) and Community and citizen stakeholders (local communities and citizens, NGOs/associations, charities, informal groups, particularly those working with vulnerable groups). Based on literature research, a preselection was made of certain hazards that were considered relevant for the areas: Earthquake; Volcanic Eruption; Tsunami; Landslide; Heatwave; Storm; Blizzard; Flood; Drought; Wildfire; Epidemics/Pandemics; Technological accident; Cyber threat; Terrorist attack; and Chemical, Biological, Radiological, Nuclear and Explosive (CBRNe) malicious act. Data collection for the survey and interviews took place between February and November 2023 using convenience sampling. Informed consent forms were signed before the interviews and integrated into 24 the questionnaire. In respect of ethics, the online survey, the interview guideline and the informed consent for participation were translated into Greek, French and German. Furthermore, most of the interviews were conducted in the native language of the interview partners. Answers were analysed in anonymous form using qualitative content analysis (Mayring, 2002) and descriptive statistics in IBM SPSS V 28.0.0.0 and Microsoft Excel 2016. An online questionnaire was created in Greek, French, and German, and the link was sent to experts active in these regions via email. A total of 71 participants responded to the online questionnaire (19 partly, 52 fully filled out; 26 were from Greece, 13 from France, and 32 from Austria). Nearly half of the respondents (49.3%) were female, and the mean age was 44.8 years (SD = 11.8). Most participants (60.6%) were working in disaster management, with a majority working for an NGO or other social organisation (28.2%), in first response (21.1%), or academia and research (16.9%). Relevant questions for social vulnerability factors aimed to collect expert opinions on socially vulnerable groups, their vulnerability in different hazard situations as well as the status of and the potential for their involvement in disaster management and education. For instance, participants were asked to identify the top five hazards in their regions (using the preselected list of hazards mentioned before), and to list the groups they considered most vulnerable to these hazards and why. In order to collect qualitative in addition to the quantitative survey data, eleven semi-structured interviews were conducted with three experts in Greece, four in France, and four in Austria. Almost half of the interview partners were female, and their age ranged between 31 and 66 years. Their expertise spanned diverse fields ranging from social sciences over work with the homeless to disaster management. The interview guideline included questions about the most relevant hazards in the interviewees' respective regions, the most relevant vulnerable groups to these hazards according to their experience, the reasons they considered them vulnerable, as well as whether they thought gender played a role in vulnerability. Seven interviews were conducted in oral form (online, via telephone, or in person) and four in written form via email. 3. Results As a result of the literature research, the expert interviews and survey, a list of 16 possible factors covering social vulnerability to a wide range of hazards was created (see Table 1). In addition to reasons for why these groups are particularly vulnerable, the table also includes possible methods of quantification or operationalization of vulnerability for a certain area or community. The list was retrospectively clustered into six categories: life-stage-, health-, social-connection-, resource-, exposure-and-protection-, and knowledge-and-awareness-related vulnerabilities. Table 1. Vulnerability indicators devised for Paris, Athens, and Vienna, in no particular order Vulnerability Indicator Reasons for vulnerability Potential measurement Life-stage-related Advanced age Lack of capacity to respond to disasters, dependency on others, mobility problems % of people over 65 years of age in the population (see e.g. Chou, 2004) Young age Lack of capacity to respond to disasters, dependency on others, mobility problems % of people under 15 years of age in the population (see e.g. Chou, 2004) 31 2. Matching concepts In the frame of this paper, we discuss three concepts to match measures and targets of pandemic management that were applied within the Austrian national security project ROADS on decision support for pandemic management (Rainer et al, 2024, ROADS, 2025): • Taxonomies • Expert knowledge • Metrics The solutions tackled in ROADS are pandemic management measures that can be used at the operational, tactical, or strategic level, depending on the selected targets. The gaps intended to be closed are represented by single targets or bundles of targets arising during the management of pandemics such as protection of the vulnerable population. 2.1 Matching using taxonomies An option for matching the needs and measures of national pandemic management is the development of a taxonomy of pandemic management functions. The matching process is very much based on the concept of the Portfolio of Solutions (PoS) (DRIVER+, 2025). In the PoS, the matching process links solutions (i.e. tools, software, processes, etc.) with one or more Crisis Management Functions (CMFs). This method was largely adapted in ROADS, with a key difference, namely • The integration of (usually detrimental) secondary effects and necessary framework conditions are presented for each measure; a comparison of measures is also possible here. Because it turned out that existing international taxonomies for crisis management as well as pandemic management were only partially suitable for the developed demonstrator, new pandemic management functions were developed to build a specific taxonomy of pandemic management functions. The principle of matching relies on using the same taxonomy of pandemic management functions (PMFs) to describe both measures and objectives. Table 1 provides selected examples from the taxonomy of pandemic functions. Table 1. Examples of elements of the taxonomy of pandemic functions from ROADS Pandemic Management Function elements Detection of pathogens Assessment of the Reproduction Rate Identification of the transmission path of diseases Source: (authors) Pandemic management functions aim to achieve effects in a pandemic response system (e.g. coordination, alignment of efforts, shared awareness). The ‘function’ focuses on what is to be achieved, not how or by whom. Individual tools, systems, building blocks, can each fulfil a specific function – either independently or in combination. Conversely, a single element may also support multiple functions. (DRIVER+ Terminology, 2025). There are different approaches on how to develop a taxonomy of pandemic management functions. In ROADS, so-called pandemic characteristics were developed based on expert knowledge and the taxonomy of pandemic management functions was derived from them. The development of pandemic management functions 32 from pandemic characteristics is demand-orientated and correspondingly ambiguous. For this reason, it is also necessary for a taxonomy to be expandable at any time and therefore, it is dynamic by nature. Elements of the taxonomy can only be reduced if it is ensured that the removed element has not been used in any description of a measure (solution) or target. The number of identical pandemic management functions assigned to both a target and a possible corresponding measure can be used to rank how well different measures address that target. 2.2 Matching based on expert knowledge In addition to algorithmic matching based on a taxonomy, a knowledge base maintained by experts can be used. Experts define specific relations between objectives and measures. Some examples of such relations are given in Table 2 for illustration purposes. These recommendations are based on scientific findings and practical experience, typically captured in workshops and interviews. However, the expert assessments need to be regularly updated to reflect new scientific findings. Table 2. Examples of target – measures relations as basis for expert based matching Strategic target Tactical target Operational target Measures Containment Reduction of morbidity & mortality Reduction of virus spread Hygienic measures: e.g. hand washing Containment Reduction of morbidity & mortality Reduction of virus spread Quarantine Protection by medical counter measures: vaccination and medication Minimising severe disease progression Contribution to development of medical countermeasures Vaccination (if available) Source: (authors, CAVE (2021)) The expert knowledge management approach has some advantages: • Greater precision through qualified assessments compared to taxonomy matching • Measures can be prioritised On the other hand, there is need for continuous maintenance of the knowledge base, in particular during the dynamic evolution of a pandemic like COVID-19, the scientific basis changed almost on a daily rate. Moreover, as with all notor only semi-automated approaches, the scalability is limited in case of manual interventions. If large volumes of new information need to be integrated in the knowledge base, experts may be hindered by limited resources. Expert systems are highly dependent on the selection of panel members and thus prone to bias. In general terms, lack of standardised definitions of the requested expert status as well as of harmonised good practices are limiting factors (see e.g. Langfeldt, 2001, Hossain et al 2018). 2.3 Metric based matching Both the taxonomyand expert knowledge-based matching approaches allow to establish qualitative relations between measures and targets. By extending the expert knowledge approach towards metricbased matching, quantitative matching relations can be established. For that purpose, it is necessary 33 to have scientific information available that provides a link between measures and metrics. In Table 3, examples are shown illustrating the potential use of metrics for matching. Table 3. Example of specific measures and associated metrices Measure Focus of measure Impact on metric Lockdown Store closure Reduction of reproduction rate by 35% (95% CI: 29-41%) Lockdown Closing of gastronomy business Reduction of reproduction rate by 12% (95% CI: 8-17%) Lockdown Leisure and entertainment facilities closure Reduction of reproduction rate by 3% (95% CI: -1-10%) Source: (Sharma et al 2021) These are examples of measure – metric relations based on a systematic literature search. Such metrics have the potential to be used for quantitative matching of e.g. the operational target to reduce the virus spread and specific measures within the categories of lockdowns. It becomes evident that this approach enables a quantifiable comparison between store closure and the closure of gastronomy business. Nevertheless, this type of matching should be based on the outcome of a large body of evidence – not a single study, encompassing aspects such as type of pathogen, regional occurrence, type of affected population, among others. A shortcoming of this approach was that only a limited number of studies were found that provided this type of information. At this stage, quantifiable indicators are used instead of elements of a taxonomy. 2.4 Alternative matching concepts The three matching concepts discussed in the previous chapters all have limitations because of their rather static data collection and aggregation processes. The decision support based on this type of non-dynamic, common operational picture aggregation hampers the optimised selection of measures to tackle the dynamic changing situation during a pandemic such as COVID-19. There is need for dynamic collection, structuring, evaluation, aggregation, and finally provision of knowledge. This encompasses AI-based allocation of measures to expected impacts. Such approaches must be grounded in a detailed evaluation of existing workflows of pandemic management, encompassing analysis of mandates and roles. It is also worth considering why vectorisation approaches were not used in the proposed approach. Their lack of transparency, challenges in reproducibility, and limited suitability for small datasets make them less appropriate for contexts requiring clear, accountable decision-making. A structured, taxonomy-based method offers a more interpretable and reliable alternative. 3. Demonstrator for pandemic knowledge management To support evidence-based decision-making in crisis contexts, a web-based demonstrator system has been developed for structured mapping between strategic goals and corresponding measures. The core of the demonstrator is implemented using Drupal 10 as a content management system, enhanced with a multi-layer architecture to ensure scalability, usability, and extensibility: • Presentation Layer: Drupal 10 with a Bootstrap-based frontend provides an accessible user interface for domain experts. 34 • Data Storage Layer: A relational database (MariaDB) is used to model structured content types for goals and measures. This choice supports well-defined relationships and maintains consistency in complex metadata structures. While NoSQL or graph databases offer benefits in highly dynamic or relationship-heavy domains, the current use case favours relational consistency and integration with Drupal’s entity system. • Search and Indexing Layer: The Apache Solr integration is a critical component of the demonstrator. Solr enables advanced full-text search, faceted filtering, and dynamic ranking of results. Content from both targets and measures is indexed along key taxonomic fields such as thematic classification, expected impact, and entry probability. This facilitates the retrieval of relevant matches in near real-time, even as the dataset grows. For example, a decisionmaker can filter all measures related to “Containment” with an “Impact = 5” and “Probability ≥ 3” within milliseconds. The Apache Solr is configured to support semantic mapping. Measures and goals are both tagged with hierarchical taxonomy terms (e.g., "Containment > Lockdown" or "Mitigation > Public Awareness"). Solr indexes these terms in flat and hierarchical forms, enabling both exact and fuzzy matching. 4. Discussion and outlook The systematic approach of matching gaps and solutions that was demonstrated both on the international level in DRIVER+ as well as in the national project ROADS has several advantages, encompassing reproducibility of the matching approach, comparability of solutions, as well as their potential to close capability gaps and the possibility to expand the matching concept based on evolving end user requirements. On the other hand, it turned out that there is need for continuous involvement of practitioners and other stakeholders to update their requirements as well as the various solutions to adapt the description of their capabilities. Moreover, when looking at the dynamic changing knowledge base in the course of a pandemic, scalability quickly reaches its limits due to frequently reaching of the maximum of available human capacities to continuously update and extend the knowledge base needed for the description of gaps, measures, as well as metrics. These limitations hamper the success of existing approaches considerably. This highlights the need to explore the potential of automated or semi-automated approaches, e.g. by AI applications to have a dynamic knowledge base available reflecting the dynamic evolution in complex scenarios such as pandemics. A web-based system was developed to support decision-makers in efficiently identifying appropriate measures based on predefined criteria. Implemented in Drupal 10, the system uses taxonomies and modules, for example, Entity Views Attachment (EVA) to dynamically visualise relationships between objectives and measures. The three matching concepts presented in this paper have specific properties: • Taxonomies enable a structured and systematic matching of measures to objectives. By developing a specific taxonomy for pandemic management functions, a clear and consistent description of both measures and objectives can be achieved. This facilitates the comparability and evaluation of measures • Expert knowledge provides a more precise and informed mapping of measures to objectives based on scientific evidence and practical experience. This method enables prioritisation of measures and takes into account the latest scientific developments, as long as the knowledge can be continuously updated • Metrics-based approaches extend the qualitative categorisation by introducing quantitative relationships between measures and objectives. This enables an objective evaluation and 35 comparability of the measures based on measurable criteria. However, this method is heavily dependent on the availability of comprehensive scientific data, which is often limited. Using the three approaches - taxonomies, expert knowledge and metrics – individually or in combination, provides a robust foundation for identifying appropriate measures for pandemic management. Each approach has its own strengths and weaknesses, and their integration has the potential to significantly improve the effectiveness and efficiency of the selection of measures. However, the dynamic nature of pandemics requires flexible and adaptable systems that can be continuously updated and expanded. Future developments should focus on the automated extraction, structuring and semantic linking of knowledge of heterogeneous information sources (e.g. text-based documents) using artificial intelligence (AI) and graph databases. First, content is extracted and converted into structured semantic units, such as “solutions”, “gaps”, or “capabilities”. These entities and their semantic relations are then mapped in a graph database, creating a flexible, relational knowledge base. Generative Large Language Models (LLMs) have demonstrated remarkable capabilities in text understanding and generation, enabling the extraction of structural knowledge from plain natural language texts (Xu et al, 2023). A central element is the automatic generation of a taxonomy, which hierarchically structures the extracted concepts along thematic and functional dimension. The semantic grouping will then be integrated into the graph structure. Next an intelligent matching engine will be developed that automatically correlates solutions with identified gaps and vice versa. The matching is based on several dimensions like semantic similarity (via embedding comparison), taxonomic proximity (same or related categories) and contextual weighting based on metrics such as relevance, effort or technical fit. Another important component will be a feedback mechanism through which user can provide feedback on the quality of the suggested matches. These evaluations are transferred into a learning process to iteratively improve the matching logic – e.g. through re-ranking models or reinforcement learning. This could further improve decision-making and enable a faster and more precise response to changing pandemic situations. Acknowledgement The research leading to these results has received funding from KIRAS Cooperative R&D Projects 2021 ROADS with project number FO999899442 as well as funding from FP7 Project DRIVER+ with project number GA No. #607798. Microsoft 365 Copilot was used to summarise and optimise parts of this paper. Part of the text of this paper was initially available from the author team in a German version and was initially translated using the free version of DeepL Translator before being adapted by the author team. References Agrawal, G.; Deng, Y.; Park, J.; Liu, H.; Chen, Y.-C. Building Knowledge Graphs from Unstructured Texts: Applications and Impact Analyses in Cybersecurity Education. Information 2022, 13, 526. https://doi.org/10.3390/info13110526 CAVE 2025, Community Engagement and Vulnerabilities in Coping with Epidemics, https://www.kiras.at/en/financedproposals/detail/projekt-cave/ retrieved 24.04.25 CMINE 2025, The Crisis Management Innovation Network Europe, https://www.cmine.eu/, retrieved 24.04.25 CORDIS 2025, EU research Results, https://cordis.europa.eu/, retrieved 24.04.25 DCNA 2025, Disaster Competence Network Austria, https://www.dcna.at/index.php/en/home.html, retrieved 24.04.25 36 DRIVER+ 2025, Driving Innovation in Crisis Management for European Resilience, https://www.driverproject.eu/driver-project/, retrieved 24.04.25 DRIVER+ Terminology, 2025, https://www.driver-project.eu/driver-project/terminology/, retrieved 24.04.25 DRSH 2025, Disaster Risk Stakeholder Hub, https://www.cmine.eu/topics/35391/page/home, retrieved 24.04.25 EPC 2025, Knowledge Hub – A comprehensive collection of online tools, resources and publications, https://www.epcresilience.com/knowledge-hub, retrieved 24.04.25 Hearst, Marti: Automatic Acquisition of Hyponyms from Large Text Corpora. Proceedings of the 14th Conference on Computational Linguistics (CoLing). 10.3115/992133.992154. 2000 Hossain, M.I., Kobourov, S., Purchase, H., Surdeanu, M, ReMatch: Research Expert Matching System, 2018 International Symposium on big Data Visual and Immersive Analytics, IEEE Xplore, doi: 10.1109/BDVA.2018.8534021, 2018, retrieved 24.04.25 Langfeldt, L: Decision-making in expert panels evaluating research – constraints, processes and bias, Dissertation, University of Oslo, https://nifu.brage.unit.no/nifu-xmlui/bitstream/handle/11250/274303/NIFUrapport20026.pdf?sequence=1, 2001, retrieved 24.04.25 PoS 2025, Portfolio of Solutions, https://www.driver-project.eu/training-module/ retrieved 24.04.25 Rainer, K., Grosser, C., Hagendorn, M., Hafner, A., Leidwein, A., Sturm, N., Geyer, C., Salomon, G., Prüller, R., Schöggl, F. From strategic planning to tackling pandemics, New challenges for ICT and management, Schriftenreihe Informatik 52, 31st Interdisciplinary Information Management Talks, Prague University of Economics and Business, DOI:10.350.11/IDIMT-2023, 2023 Rainer, K., Kundratitz, V., Hagendorn, M., Neubauer, G., Ignjatovic, D., Kutalek, R., Sturm, N. Artificial Intelligence in Crisis Management: Potential Solutions and Challenges, Schriftenreihe Informatik 53, 32nd Interdisciplinary Information Management Talks, Prague University of Economics and Business, DOI:10.350.11/IDIMT-2024, 2024 ROADS 2025, ROADS to Health: Decision support for the Pandemic Management of the Future, https://www.kiras.at/en/financed-proposals/detail/roads-to-health-decision-support-for-the-pandemicmanagement-of-the-future/ Rotmensch, Maya & Halpern, Yoni & Tlimat, Abdulhakim & Horng, Steven & Sontag, David:. Learning a Health Knowledge Graph from Electronic Medical Records. Scientific Reports. 7. 10.1038/s41598-017-05778-z., 2017 Roy, S., Ghosh, N., Uplavikar, N., Ghosh, P. Towards a Unified Pandemic Management Architecture: Survey, Challenges and Future Directions, arXiv:2022.07448, Cornell University, https://arxiv.org/abs/2202.07448, retrieved 24.04.25 Sharma, M., Mindermann, S., Roger-Smith, C., Leech, G., Snodin, B., Ahuja, J., Sandbrink, J.B., Monrad, J.T., Altman, G., Dhaliwal, G., Finnveden, L., Norman, A.J., Oehm, S.B., Sandkühler, J.F., Airchison, L., Gavenciak, , T., Mellan, T., Kulveit, J., Chindelevitch, L., Flaxman, S., Gal, Y., Mishra, S, Bhatt, S., Brauner, J.M. Understanding the effectiveness of government interventions against the resurgence of COVID-19 in Europe, NATURE COMMUNICATIONS,12:5820,https://doi.org/10.1038/s41467-021-26013-4, www.nature.com/naturecommunications, 2021 Tagarev, T.; Ratchev, V. A taxonomy of crisis management functions. Sustainability 2020, 12, 5147. Xu, D., Chen, W., Peng, W., et al: Large Language Models for Generative Information Extraction: A Survey. arXiv:2312.17617, 2023 37 THE USE OF AN AI-SUPPORTED TOOL FOR THE DEPLOYMENT OF INFORMAL VOLUNTEERS IN CRISIS AND DISASTER MANAGEMENT ON THE BASIS OF THEIR COMPETENCIES Sabine Kretschy, Nadine Sturm, Martin Söllner Johanniter Österreich Ausbildung und Forschung gem. GmbH [email protected] Christoph Angster Federal Ministry Labour, Social Affairs, Health, Care and Consumer Protection, Republic Austria [email protected] Andreas Rath ONDEWO GmbH [email protected] Bernhard Bürger Austrian Institute of Technology GmbH [email protected] David Schneeberger Research Institute AG & Co KG [email protected] Georg Hahn OSSBIG Austria [email protected] Sophie Zeissler NOUS Wissensmanagement FlexCo s.zeiss[email protected] 38 Birgit Pröll Johannes Kepler University Linz [email protected].at Sandra Pichler Disaster Competence Network Austria [email protected] DOI: 10.35011/IDIMT-2025-37 Keywords Informal volunteers, disaster/crisis management, AI, chatbot, skills/competencies Abstract Volunteers play an important role in crisis and disaster management in Austria. However, the competence-based integration of informal, spontaneous volunteers in crisis situations remains a major challenge for public safety organisations. The CERTIFIER (Certificate for CERTificate For Integrated Emergency Response) project addresses this issue by developing a digital, competencybased system, providing a basis to efficiently and securely integrate informal volunteers into emergency operations. The core innovation of the project is a digital proof of competence system that allows individuals to present verifiable qualifications through interoperable digital credentials. CERTIFIER allows targeted deployment based on validated skills, supported by ontology-based competence matching and a user-friendly chatbot interface. A mixed-methods requirements elicitation and tabletop-exercises showed a strong interest in automated registration, documentation of skills and certification of engagement. The results highlight the potential of the system to optimise human resource allocation and improve operational efficiency. Key design considerations support the registration process and the development of an ontology for matching skills and competences, as well as transparent handling of data. CERTIFIER addresses current gaps in coordination and resource allocation and demonstrates how digital and AI technologies can promote responsive and effective volunteer engagement in disaster management. 1. Background and Introduction Volunteering is one main pillar of crisis and disaster management in Austria, with emergency response organisations ensuring the effective and professional handling of emergency and disaster operations (BMI, 2009). Organisations managing disasters are used to working with trained, formal and professionalised volunteers. Recent findings from the CIVolunteer (KIRAS, AT) research project highlight a trend to shortcomings in the volunteer sector: a decline or shortage of volunteers could lead to supply bottlenecks and make effective coordination in crises difficult (Schönböck et al., 2024a, Schönböck et al. 2024b, Fritz & Mathewson, 1957; Quarantelli & Dynes, 1977; Skar et al., 2016; Strandh & Eklund, 2018). However, the coordination of spontaneous, informal volunteers, who spontaneously offer support in a crisis situation is a major challenge for authorities and public safety organisations. Emergency services are also confronted with informal volunteers who form self- 39 organised groups and take actions (Quarantelli & Dynes, 1977; Stallings & Quarantelli, 1985; Strandh & Eklund, 2018). These persons or groups can provide valuable resources and make important contributions to disaster management (Barsky et al., 2007; Helshloot & Ruitenberg, 2004; Scanlon et al., 2014; Whittaker et al., 2015). They offer benefits such as reduced bureaucracy, greater flexibility and closer proximity to those affected (Stallings & Quarantelli, 1985; Whittaker et al., 2015). When it comes to managing a collaborative emergency response, public safety organisations face the challenge of how to appropriately involve informal volunteers. These actors typically do not use standardised systems and are only loosely integrated into official coordination structures. Emergency response organisations are challenged by the situation to use the potential of informal volunteers while addressing concerns that often prevent their full integration. These concerns include safety risks due to insufficient information, lack of protective equipment, limited risk awareness (Harris et al., 2017), logistical challenges, overstretch of critical resources (Harris et al., 2017; Whittaker et al., 2015), inadequate planning (Skar et al., 2016) and legal uncertainties such as liability issues (Barsky et al., 2007). This results in informal volunteers being involved only in minor tasks, while their specific skills and competencies remain unused. This is where the KIRAS funded project CERTIFIER (FFG, AT, 2023-2025) comes in. The aim of the project is to develop a comprehensive concept for the competence-based involvement of informal volunteers. A central objective is the creation of a digital solution to record and document the skills and qualifications of informal volunteers through a digital competence certificate, integrating a formal confirmation of their activities as foreseen in national law by using the digital volunteer passport, provided by the Federal Ministry of Labour, Social Affairs, Health, Care and Consumer Protection in Austria. Overall, CERTIFIER offers the following added value: • Documentation and validation of competences of informal volunteers • Basis for efficient and targeted involvement of informal volunteers in acute crisis situations • Needs-driven technology development to ensure greater stakeholder acceptance • In-depth understanding of the challenges associated with the participation of informal volunteers 2. Concept and Approach 2.1 CERTIFIER-Concept The proof of concept developed in the CERTIFIER project enables informal volunteers with existing skills to identify their qualifications to emergency services without compromising privacy or data security. This allows emergency services to make informed decisions for involvement of individuals in disaster management tasks by relying on a validated digital record of volunteers’ competencies, thereby avoiding lengthy verification procedures. This approach helps to manage disasters more effectively and with greater resource efficiency. CERTIFIER supports disaster management in the following ways: • Volunteers register via (diverse) volunteer platforms or using the AI-driven chatbot of CERTIFIER and confirm their general willingness to support tasks. Existing qualifications will be listed on the platform and can be validated by digital certificates • Volunteers, which hold digital certificates in interoperable formats (e.g. European Digital Identity, ISO/IEC 18013.5, W3C Verifiable Credential Model) received during 40 education and training (e.g. professional qualifications, driving licences, etc.) are able to share them easily using the CERTIFIER platform. • In the back end of the system, tasks are matched to their required competences, and crisis managers are informed about the available volunteer resources. • Platform operators from emergency services can notify relevant volunteers with the required skill profiles. If special qualifications are required, the chatbot serves as an interface for verifying credentials. In line with security and privacy-by-design principles, volunteers use CERTIFIER to prove that they hold the required digital credentials. The chatbot provides information about the deployment and potential areas of activity for volunteers to support emergency services. If certain qualifications are missing, the system can suggest alternative suitable fields of activity. If the validation was successful, volunteers receive a signed token (e.g. visualized in the form of a QR code) confirming their qualification and active registration. This token can be verified, e.g. through an app-based scan, by emergency services and grants permission to access the disaster area and perform specific tasks on site. CERTIFIER supports emergency services in the targeted and demand-driven allocation of volunteers. Commanders can search for specific tasks (e.g. "set up lighting") and receive AI-based suggestions of verified, qualified volunteers (e.g. electricians). Final task assignments always remain with human decision-makers. Key Innovations of CERTIFIER: • AI-supported registration of spontaneous volunteers with validated skills — privacyfriendly through verifiable credentials (SSI). • Immediate access to relevant competencies — no manual checks required. • Compatible with established digital ID standards (e.g. EUDI, W3C VCs, ISO/IEC 18013.5). • Intelligent backend enables competence-based task allocation in real time. • Volunteers retain control over their digital identity while enabling coordination transparency. • Modular and interoperable — adaptable to various disaster types and organizations. • Early registration boosts preparedness, resilience, and sustainable volunteer engagement. In case of power outages, token authenticity can still be verified offline; with internet access, additional information becomes available. 47 Figure 1. PRISMA diagram Source: authors according to PRISMA (2021) First, all papers that are not available in “Open Access” have been excluded. This criterion was included on the grounds that the scientific debate identified is mainly to help future research identify the main tools for using AI within corporate practice. Another exclusion criterion was the year of publication of the paper. As this is an assessment of the current situation, all papers with a publication date before 2023 were excluded. Besides the fact that AI is constantly evolving and improving, so keeping an eye on the current literature is necessary (Grewal et al., 2024), other research on this topic also used a 3-year time period (Woo & Choi, 2021). Other researchers, like Zawacki-Richter et al. (2019), show that the limitation of the time period can be due to some significant change, invention, or technology introduced. Therefore, the year 2023 was chosen as the starting point because it introduced the generative AI model ChatGPT for public use (Thompson et al., 2023), which is the most famous and well-known AI of current time. Subsequently, all papers that are not in the “Article” or “Proceeding Paper” category have been excluded. This step ensures a higher scientific relevance of the articles. Finally, those contributions that are not written in English have been excluded. This criterion was applied because English is the main language used in scientific publications (Bahji et al., 2023) and analyses are often made on for English written papers, as reported by Mishra et al. (2024). Subsequently, a reduction was made in terms of specific areas. The contribution aims to evaluate the given topic in business practice. Therefore, several areas within the Web of Science Categories were selected, namely “Business”, “Management”, “Operations Research Management Science” and “Computer Science Artificial Intelligence.” This step ensured a targeted focus of the research on businesses and their practices. To ensure coverage of the business management field, the “Research Area” specification was used, namely “Business Economics”, “Operations Research Management Science”, “Public Administration”, and “Engineering”. After applying all the specified criteria, 82 contributions were found. Subsequently, a manual screening of the remaining contributions was carried out, during which abstracts, titles, and keywords were analysed. The final number of papers was 66. These papers were further analysed in VOSviewer. This tool enables the construction and visualisation of bibliometric networks, such as co-authorship, keyword co-occurrence, and country collaboration maps. The tool is widely used in bibliometric research (van Eck & Waltman, 2010). 48 3. Results The first result of the analysis was finding the countries where the papers were published. It was based on 66 articles that emerged from the methodological part of the research. The analysis shows that countries such as the US, the UK, and China have a significant presence. This topic is also addressed in Germany, France, and Australia. This diverse geographical distribution underpins the fact that AI and crisis management are global topics. Second result of the analysis was keyword analysis, which was also prepared in VOSviewer, and was based again on 66 articles. The results show that the combination of AI and crisis management focuses not only on the technological aspect but also on supply chains, communication, or risk. Several main headings were identified, as shown in Figure 2. The first strand is AI and machine learning - these are technologies closely related to crisis management. The second strand is then resilience within crisis management - a clear focus on emergencies, their management, and resolution. The third strand is then forecasting and data analytics - an area defining decision support tools. This is followed by an area focused on organisational aspects - showing the broader context of crisis management within the supply and demand chain. The final strand is then major crises such as the Covid-19 pandemic. Figure 2. Keyword analysis Source: authors A deeper keyword analysis identified three main thematic clusters that represent the most important directions of current research on the link between artificial intelligence and crisis management. Each cluster has its own logic and practical implications. The first cluster focuses on technologies and methods applied in crisis management. This cluster is built around artificial intelligence, machine learning, deep learning, neural networks, optimization, prediction, modelling, big data, and data analytics. These terms form the technological backbone of the entire topic under study. In practice, much of the current research focuses on developing and using algorithms capable of predicting the occurrence of crises, simulating possible scenarios, analysing large volumes of data from sensors or social networks, and optimising decision-making under time pressure. A typical example is using machine learning models to predict the spread of natural disasters such as earthquakes, floods, or pandemics or to optimize logistics routes for humanitarian aid. A specific example can be envisioned as AI models predicting flood occurrence based on meteorological and subsequent use of algorithms to optimize humanitarian aid routes in real-time. 49 The second cluster focuses on crisis management itself and on building resilience. Keywords such as crisis management, disaster management, disaster relief operations, resilience, coordination, and crisis communication show that this area deals with the practical management of crises. Research in this area focuses on how to structure decision-making processes in times of crisis, how to effectively coordinate multiple actors (e.g., government agencies, non-profit organisations, businesses), how to ensure fluid communication in chaotic environments, and how to systematically strengthen the ability of organisations to cope with unpredictable events. The practical implication of this cluster is that AI systems are used not only to detect crises but also to directly support the management of crisis teams, optimize interventions on the ground, and model future scenarios to minimize damage. This area can consider a few specific applications and situations, such as using AI systems to manage incidents during cyber-attacks, testing the resilience of operations through AI simulations of personnel outages, or simulating the impact of a supply chain outage and designing alternatives. The third cluster focuses on the social and communication dimensions of crisis management. It includes keywords such as crisis communication, social media, information, or behaviour. This cluster reflects that crisis management is impossible without effective information management in today's digital age. The research here explores, for example, how social networks influence the spread of information and misinformation during crises, what communication strategies are most effective in alerting the public, and how social media data analysis can be used to identify problems quickly. In practical terms, this means that AI is being used to process large volumes of data and analyse public sentiment, monitor crisis responses, and support decision-making based on how public opinion is evolving in real-time. Again, several specific use cases can be considered, such as using social media analytics to disseminate crisis information, automating the sending of personalized alerts to the population via chatbots, or monitoring customer reaction when products are unavailable. 3. Discussion The first finding was that the link between AI and crisis management lies in technology. AI is revolutionizing crisis management through various technological applications. It improves decisionmaking, information management, and communication during crises. Key technologies include machine learning, social network analysis, big data analytics, and geographic information systems. However, challenges remain, including managing disparate information, supporting different platforms, or data set limitations (Bukar et al., 2022). Artificial intelligence applications, such as drone technology or robotics, contribute to more effective disaster response and mitigation. Overall, the role of AI in crisis management is crucial for faster responses in all phases of disaster management (Arfan et al., 2019). The above facts support the finding that it is necessary to emphasize this issue and help businesses overcome this obstacle. Subsequently, the area of crisis management itself was observed, with the concepts of resilience, communication, or coordination identified as important. Crisis management and resilience are crucial for organisations facing major disruptions. Effective crisis response involves shared sensemaking, issue identification, and strategic adaptation (Parker, 2023). Organisational resilience is characterized by preparedness, response effectiveness, and recovery capabilities (Almufarji & Husin, 2022). Building long-term organisational resilience requires integrating crisis management and resiliency theories and embedding innovations introduced during emergencies (Donelli et al., 2022). Finally, crisis communication was identified as an important area. Crisis communication has emerged as a crucial aspect of crisis management in the digital age, with social media playing a pivotal role (Kuipers et al., 2023). Effective communication and social media usage are the most significant factors influencing crisis management, followed by leadership and knowledge management (Hazaa 50 et al., 2021). Future research should focus on leveraging AI-based technologies to improve crisis communication practices and establish shared situational awareness (Abboodi et al., 2023). To safely use AI in crisis management, the necessary ethical and legal standards must first be in place. These will establish accountability for AI systems' decision-making (Gevaert et al., 2021). Ghaffarian et al. (2023) then add that openness is crucial to ensure interpretability so that AI decisions can be tracked and explained. Equally important is the collaboration between government institutions and companies, and hence research organisations, which will support the development of reliable systems while enabling rapid real-time response (Kapucu et al., 2024). Although the paper was intended to emphasize timeliness, it is important to consider the research barriers and limitations. There has been a focus on sources available in English, which reflects a desire for practical applicability to a broader professional and corporate audience, but at the same time, may lead to, but may also lead to, the omission of some specific approaches published in other languages or regions. Similarly, the choice of a timeframe of 2023 onwards responds to the dynamic developments in the field but may have overlooked older yet relevant studies. Finally, bibliometric analysis should be seen as a tool for identifying significant trends and directions, not as a means of deep content analysis of individual publications. However, these aspects do not diminish the value of the findings; on the contrary, they provide a solid basis for follow-up research and further refinement. Further research may focus on a deeper analysis of AI's effectiveness in crisis management, for example, in the form of comparative studies or other systematic reviews. There is also scope for looking at AI's role in decision-making processes or coordination of actors during a crisis. The role of AI in information and communication flows within organisations can also be explored. 4. Conclusion The main aim of this paper was to provide a bibliometric analysis with a focus on identifying the main topics in the field of “AI in crisis management”. This study has shown that (AI) plays a key role in modern crisis management and significantly influences how organisations respond to emergencies. The literature analysis revealed three main areas of interest - technology tools, building organisational resilience, and crisis communication, all complementary and reinforcing. AI enables faster threat detection, more effective decision-making, and better coordination of crisis interventions. At the same time, however, several challenges have been identified that hinder the full implementation of AI in crisis management. These include the limited availability of quality data, ethical issues, problems with integrating technology into existing structures, and the need for greater awareness of AI's potential among managers, both for managers of private companies and those working in the public sphere, as crisis management affects both spheres. In conclusion, the future success of AI in crisis management depends not only on technological advances but also on the ability of organisations to systematically incorporate new approaches and strengthen their ability to withstand crises. In addition to managers of private and public enterprises who can use the results to optimize decisionmaking processes, plan scenarios, or improve communication, the study results also have important implications for the academic and research community working to develop theoretical frameworks and practical applications of AI in crisis management. The bibliometric analysis provides an overview of current trends, research directions, and potential areas for further study. From the perspective of technology companies or developers, the results of this study provide a source of inspiration for the development of AI-based tools that will reflect the specific needs of practice. 51 References Abdalla, M., Alarabi, L., & Hendawi, A. (2021). Crisis Management Art from the Risks to the Control: A Review of Methods and Directions. Information, 12(1). https://doi.org/10.3390/info12010018 Abboodi, B., Pileggi, S. F., & Bharathy, G. (2023). Social Networks in Crisis Management: A Literature Review to Address the Criticality of the Challenge. Encyclopedia, 3(3), 1157-1177. https://doi.org/10.3390/encyclopedia3030084 Almufarji, M., & Husin, N. A. (2022). The Characteristics of Resilient Organizations within Crisis Management. Archives of Business Research, 10(2), 1-14. https://doi.org/10.14738/abr.102.11708 AlRyalat, S. A. S., Malkawi, L. W., & Momani, S. M. (2019). Comparing Bibliometric Analysis Using PubMed, Scopus, and Web of Science Databases. Journal of Visualized Experiments, (152). https://doi.org/10.3791/58494 Arfan, M., Khan, Z., Qadri, N., Hameed, M. H., & Amir, A. R. (2019). Role of Artificial Intelligence (AI) in Combined Disaster Management. Organization Theory Review, 3(2), 97-121. https://doi.org/10.32350/OTR.0302.05 Bahji, A., Acion, L., Laslett, A. -M., & Adinoff, B. (2023). Exclusion of the non-English-speaking world from the scientific literature: Recommendations for change for addiction journals and publishers. Nordic Studies on Alcohol and Drugs, 40(1), 6-13. https://doi.org/10.1177/14550725221102227 Brennan, S. E., & Munn, Z. (2021). PRISMA 2020: a reporting guideline for the next generation of systematic reviews. JBI Evidence Synthesis, 19(5), 906-908. https://doi.org/10.11124/JBIES-21-00112 Bukar, U. A., Sidi, F., Jabar, M. A., Nor, R. N. H., Abdullah, S., Ishak, I., Alabadla, M., & Alkhalifah, A. (2022). How Advanced Technological Approaches Are Reshaping Sustainable Social Media Crisis Management and Communication: A Systematic Review. Sustainability, 14(10). https://doi.org/10.3390/su14105854 Cantrell, A., Booth, A., & Chambers, D. (2022). A systematic review case study of urgent and emergency care configuration found citation searching of Web of Science and Google Scholar of similar value, 41(2), 166-181. https://doi.org/10.1111/hir.12428 Cao, L. (2023). AI and data science for smart emergency, crisis and disaster resilience. International Journal of Data Science and Analytics, 15(3), 231-246. https://doi.org/10.1007/s41060-023-00393-w Ciekanowski, Z., Nowicka, J., Zurawski, S., Czech, A., & Zdunek, M. (2023). Crisis Management and Crisis Situation in the Organization. European Research Studies Journal, XXVI(4), 339-353. https://doi.org/10.35808/ersj/3288 Donelli, C. C., Fanelli, S., Zangrandi, A., & Elefanti, M. (2022). Disruptive crisis management: lessons from managing a hospital during the COVID-19 pandemic. Management Decision, 60(13), 66-91. https://doi.org/10.1108/MD-022021-0279 Gevaert, C. M., Carman, M., Rosman, B., Georgiadou, Y., & Soden, R. (2021). Fairness and accountability of AI in disaster risk management: Opportunities and challenges. Patterns, 2(11). https://doi.org/10.1016/j.patter.2021.100363 Ghaffarian, S., Taghikhah, F. R., & Maier, H. R. (2023). Explainable artificial intelligence in disaster risk management: Achievements and prospective futures. International Journal of Disaster Risk Reduction, 98. https://doi.org/10.1016/j.ijdrr.2023.104123 Greener, S. (2022). Evaluating literature with bibliometrics. Interactive Learning Environments, 30(7), 1168-1169. https://doi.org/10.1080/10494820.2022.2118463 Grewal, D., Guha, A., & Becker, M. (2024). AI is Changing the World: For Better or for Worse? Journal of Macromarketing, 44(4), 870-882. https://doi.org/10.1177/02761467241254450 Hassan, W., & Duarte, A. E. (2024). Bibliometric analysis: A few suggestions. Current Problems in Cardiology, 49(8). https://doi.org/10.1016/j.cpcardiol.2024.102640 Hazaa, Y. M. H., Almaqtari, F. A., Al-Swidi, A., & Tan, A. W. K. (2021). Factors Influencing Crisis Management: A systematic review and synthesis for future research. Cogent Business & Management, 8(1). https://doi.org/10.1080/23311975.2021.1878979 Kapucu, N., Parkin, A., Lumb, M., & Dippy, R. (2024). Crisis coordination in complex intergovernmental systems: The case of Australia. Public Administration Review, 84(3), 389-399. https://doi.org/10.1111/puar.13830 52 Karinshak, E. (2024). Simulations in human‐AI crisis management systems. Journal of Contingencies and Crisis Management, 32(1). https://doi.org/10.1111/1468-5973.12525 Kuipers, S., Perlstein, S., Wolbers, J., & Jong, W. (2023). Assist or accuse? Identifying trends in crisis communication through a bibliometric literature review. Risk, Hazards & Crisis in Public Policy, 14(4), 272-296. https://doi.org/10.1002/rhc3.12283 Kulkarni, C. (2024). Transformation of Agriculture through Artificial Intelligence: A Comprehensive Review. Journal of Scientific Research and Reports, 30(12), 798-803. https://doi.org/10.9734/jsrr/2024/v30i122722 Mian, S. (2022). Foundations of Artificial Intelligence and Applications. Journal of Artificial Intelligence and Technology, 2(1), 1-2. https://doi.org/10.37965/jait.2022.01 Mishra, M., Desul, S., Santos, C. A. G., Mishra, S. K., Kamal, A. H. M., Goswami, S., Kalumba, A. M., Biswal, R., da Silva, R. M., dos Santos, C. A. C., & Baral, K. (2024). A bibliometric analysis of sustainable development goals (SDGs): a review of progress, challenges, and opportunities. Environment, Development and Sustainability, 26(5), 11101-11143. https://doi.org/10.1007/s10668-023-03225-w Mukhamediev, R. I., Popova, Y., Kuchin, Y., Zaitseva, E., Kalimoldayev, A., Symagulov, A., Levashenko, V., Abdoldina, F., Gopejenko, V., Yakunin, K., Muhamedijeva, E., & Yelis, M. (2022). Review of Artificial Intelligence and Machine Learning Technologies: Classification, Restrictions, Opportunities and Challenges. Mathematics, 10(15). https://doi.org/10.3390/math10152552 Nteka, N. (2021). Crises Analysis and Management. Entrepreneurship, 9(1), 64-77. https://doi.org/10.37708/ ep.swu.v9i1.6 Öztürk, O., Kocaman, R., & Kanbach, D. K. (2024). How to design bibliometric research: an overview and a framework proposal. Review of Managerial Science, 18(11), 3333-3361. https://doi.org/10.1007/s11846-024-00738-0 Pang, G. (2022). Artificial Intelligence for Natural Disaster Management. IEEE Intelligent Systems, 37(6), 3-6. https://doi.org/10.1109/MIS.2022.3220061 Parker, L. D. (2023). Third sector crisis management and resilience: Reflections and directions. Financial Accountability & Management, 40(3), 326-343. https://doi.org/10.1111/faam.12379 PRISMA Statement: Flow Diagram. (2021). Prisma-statement.org. Retrieved April 20, 2025, from https://prismastatement.org/prismastatement/flowdiagram.aspx Shrivastava, D. A. (2024). Artificial Intelligence (AI): Evolution, Methodologies, and Applications. International Journal for Research in Applied Science and Engineering Technology, 12(4), 5501-5505. https://doi.org/10.22214/ijraset.2024.61241 Singh, V. K., Singh, P., Karmakar, M., Leta, J., & Mayr, P. (2021). The journal coverage of Web of Science, Scopus and Dimensions: A comparative analysis. Scientometrics, 126(6), 5113-5142. https://doi.org/10.1007/s11192-02103948-5 Sohrabi, C., Franchi, T., Mathew, G., Kerwan, A., Nicola, M., Griffin, M., Agha, M., & Agha, R. (2021). PRISMA 2020 statement: What's new and the importance of reporting guidelines. International Journal of Surgery, 88. https://doi.org/10.1016/j.ijsu.2021.105918 Thompson, K., Corrin, L., & Lodge, J. M. (2023). AI in tertiary education: progress on research and practice. Australasian Journal of Educational Technology, 39(5), 1-7. https://doi.org/10.14742/ajet.9251 Tiwari, R. (2024). Crisis Management in Hospitality. International Journal of Advanced Research, 12(05), 850-862. https://doi.org/10.21474/IJAR01/18793 van Eck, N. J., & Waltman, L. (2010). Software survey: VOSviewer, a computer program for bibliometric mapping. Scientometrics, 84(2), 523-538. https://doi.org/10.1007/s11192-009-0146-3 Vladu, O. -M. (2023). Digital Transformation in Crisis Management: The Key Role of Artificial Intelligence. Scientific Research and Education in the Air Force, 24, 103-110. https://doi.org/10.19062/2247-3173.2023.24.15 Wong, Y. K. (2021). Dealing Crisis Management using AI. Advances in Machine Learning, Data Mining and Computing, 95-102. https://doi.org/10.5121/csit.2021.111409 Woo, J. H., & Choi, H. (2021). Systematic Review for AI-based Language Learning Tools. Journal of Digital Contents Society, 22(11), 1783-1792. https://doi.org/10.9728/dcs.2021.22.11.1783 CYBER SECURITY VS AI 55 A SECURE SCHEME FOR CHAINED AUTHENTICATION COMBINED WITH ATTESTATION Michael Sonntag Johannes Kepler University Linz [email protected] DOI: 10.35011/IDIMT-2025-55 Keywords Authentication, microservices, attestation Abstract It may be desirable to split computation into several parts and perform them on remote systems, see e.g. microservices. However, not always are these services directly and publicly available. For instance, if the calculation is performed at home, no public IP address may be available. Or the exact location of the computation should remain hidden. In these cases, the remote party may be identified solely as a Tor onion service or some anonymous cloud provider. It is then desirable to pass on the work only if the recipient can still be identified as trustworthy, e.g. by remote attestation (guaranteeing a specific calculation via the actual executable and its secure environment). Even if the location and identity of the operator is known, this is very desirable. Moreover, each entity in a chain of calculation should remain oblivious of all non-directly connected steps and any authentication information for those, including any additional data or the exact work instructions. This paper proposes a scheme to pass on work packages in a secure manner to a verified next step, while keeping all non-directly involved data hidden. 1. Introduction If intelligent agents, e.g. AI systems, cannot perform some task immediately, they may decide on whom to pass it to for further processing. The same obviously is possible manually. In many cases the actual counterpart may be of little interest, and e.g. be selected solely based on the price: it is “merely” necessary that a specific computation is performed. This might require identification of the owner so that the work performed can be invoiced (deducted from the “points” in an account, billed in some currency etc). If this happens several times in a sequence it is necessary that e.g. a task does know whom to contact next and how to securely identify it (or at least verify their computation; see below), but none of the information used for payment – else it could submit other tasks under the identity of someone else who would be billed for it – or simply steal that data and use it a second time/somewhere else. The scheme presented here draws on the ideas implemented in Tor, specifically layered encryption. 56 1.1 Related work A somewhat similar approach is described in (Sollins, 1988) as “Cascaded authentication” but there no attestation is included and it work more as a delegation of rights. It is based on a “passport” signed at each stage, which includes limits each stage may perform on behalf of the issuer. The system described by (Bela/Piroska 2009) consists of a central “coordinator” node, which shares passwords with each node. However, if many requests come in, the load on this central node might be too large, so a stage may authenticate to another stage before passing the work on, i.e. stage i assures stage i+1 that it did validate the origin. Therefore, a trust relationship between all those stages in required (and attestation is not included either). Attempting to achieve less authentication than documentation is the scheme proposed by (Pattaranantakul et al 2021). Here the way a packet traverses multiple stages of work (e.g. firewall, load balancer, NAT) is verified through chained signatures. I.e., there the path is verifiable, but no authentication takes place and no data is kept secret. 1.2 Usage scenario When training an AI system, outsourcing calculation is often unavoidable, especially for additional company-specific training, as the necessary specialized hardware will neither be available nor costeffective to purchase. Additionally, several stages may be required, e.g. pretraining, adding preand/or post-filters, combining it with other systems and libraries, integration into a web-system (e.g. chatbot), testing & publishing it etc. Each stage obviously needs its own information to be able to perform its work. However, it does not need information and instructions for any other steps. Instructions for previous and the current stage can obviously be removed before passing the work package onward, but those for latter ones must invariably be present. Similar, but not identical, authentication information for the next stage is required. For the next stage it must be available in a “useful” format so authentication can succeed, but further stages should remain unknown and any other authentication data unavailable. Note that this approach assumes a direct hand-off between the stages (→ sequence model), i.e. not returning it to the initiator (→star model) after each step. This is especially important in AI, as models may be huge and transfer bandwidth and time is an issue – less between professional services but much more so with a single central hub at a private company. In this scenario attestation is slightly less important, as each stage is likely known in advance. But small price differences per hour for training might result in a large end prices; cheaper alternatives may be useful. If the operator of a stage is not perfectly trustworthy (perhaps because until then little known/no previous experience/…), it is important to prove the actual computation performed. Note that especially with AI models it may be practically impossible to determine whether a model was trained correctly or not afterwards – the result could mostly perform as expected but still fail for some inputs, because not all input data provided was used in training or additional malicious input was inserted. So process attestation through the software executed is then useful. A prime example for attestation being imperative is, if the operator of a stage remains anonymous and the authentication information includes some valuable consideration (directly, pre-paid etc). That information should only be disclosed to a stage if that stage can prove that it actually will perform the work and do this completely as advertised. Note that while we cannot prevent a DoS attack by e.g. carefully timed crashing/powering off the system, the system as a whole (OS, libraries, calculation program etc) cannot be modified maliciously or attestation will fail. 63 AI, CYBERSECURITY AND THE REAL VALUE PROPOSITION FOR BUSINESS: MOVING BEYOND THE HYPE Martin Zbořil Prague University of Economics and Business Faculty of Informatics and Statistics [email protected] DOI: 10.35011/IDIMT-2025-63 Keywords AI, Cybersecurity, Value Proposition, Business, Marketing Abstract In today's digital world, cybersecurity is a major concern for businesses everywhere. This paper looks at how artificial intelligence (AI) can help improve cybersecurity measures in nowadays organizations. Through a systematic review, the author identified eight particular use cases where cybersecurity might benefit from AI. Among these use cases belong e.g. enhanced threat detection, automated incident response, user behavior analysis, or application security. By combining AI with human expertise, businesses can better handle cybersecurity challenges and make smarter decisions. While AI has great potential, the technology is often hyped up, being often rather a buzzword in vendor’s marketing campaigns. This vendors’ pressure then leads to unrealistic expectations of organizations. The goal is to provide a systematic view of AI's role in cybersecurity, helping companies use its strengths while being aware of its limitations. 1. Introduction In today's digital age, cybersecurity has become a significant concern for organizations across all sectors. The escalating reliance on digital infrastructure, coupled with the increasing sophistication of cyber threats, has made traditional security measures insufficient. Consequently, organizations are driven to explore innovative solutions to protect their assets and operations. Among the emerging technologies, artificial intelligence (AI) stands out for its potential to fortify cybersecurity defenses (Raval et al., 2024). AI components, such as machine learning, natural language processing, and predictive analytics, promise more efficient and effective cybersecurity mechanisms by enabling proactive threat detection and automated response systems. However, the cybersecurity industry has seen inaccuracy in AI-branded solutions, with vendors often overstating the capabilities and benefits of their products (Content Engine LLC, 2023). This trend has led to AI being frequently used as a marketing buzzword, obscuring the actual value these technologies bring. Organizations must approach AI-driven cybersecurity solutions with caution, ensuring they understand the underlying technologies and their relevance to their specific contexts. 64 The effectiveness of AI in cybersecurity depends on several factors, including the quality of data used for training algorithms and the sophistication of the AI models (ENISA, 2023). This paper presents the initial findings of a long-term research initiative focused on the materialization of cybersecurity value for businesses. Given AI's significant influence across organizations and various domains, the starting point of this research is dedicated to the use of AI, specifically identifying security risks associated with deploying AI solutions within organizations. Through extensive research of published studies, the author has already identified eight primary security concerns (Zbořil, 2024): 1. Adversarial attacks - AI models can be manipulated through adversarial attacks, compromising their reliability and safety in critical applications (Raval, 2024). 2. Insufficient data governance leading to data leakage - Poor data governance can result in data leakage, exposing sensitive information and undermining data privacy (Kalodanis, 2023). 3. Data privacy and integrity: Unauthorized access and manipulation of data used by AI systems can lead to privacy breaches and biased outcomes (Nigro, 2023). 4. Model theft and reverse engineering: Unauthorized replication or analysis of AI models can lead to intellectual property theft and expose security vulnerabilities (Mirsky, 2023). 5. Bias and fairness: AI systems may be affected by biases from training data, leading to unfair or discriminatory outcomes (Jia, 2022). 6. Lack of explainability and transparency: The "black box" nature of AI can limit understanding and trust in AI-driven decisions (ENISA, 2023). 7. Compliance and regulatory risks: AI deployments must comply with evolving regulations to avoid legal penalties and operational disruptions (Nature, 2023). 8. Robustness and physical harm: AI systems must be robust to prevent failures that could result in physical harm or financial losses (Zaman et al., 2021). Organizations must incorporate these risks into their decision-making processes regarding the adoption and implementation of AI. Failure to consider these potential risks exposes organizations to potential business disruptions (Nigmatov & Pradeep, 2023). Nevertheless, AI is simultaneously highlighted for its potential to enhance cybersecurity mechanisms. To realize the tangible benefits of AI in cybersecurity, organizations must adopt a balanced perspective, recognizing both the potential and limitations of these technologies. By investing in robust data governance frameworks, adversarial training, explainable AI research, and continuous compliance monitoring, organizations can enhance the security posture of their AI-driven cybersecurity solutions (Hannecke, 2023). This approach enables businesses to make informed decisions regarding the integration of AI into their security strategies, moving beyond the hype to leverage AI's true value in protecting against evolving cyber threats. This paper critically reviews the usage of AI in cybersecurity, identifying areas where AI can contribute to enhancing security postures while also highlighting the limitations and risks associated with its deployment. By adopting a balanced perspective, this research aims to provide insights that enable businesses to make more informed decisions regarding the integration of AI into their cybersecurity strategies. 65 2. Methodology This research employs a comprehensive literature review methodology, including analysis of academic studies, industry reports and case studies, with the objective to synthesize current knowledge on the real security benefits of integrating AI into cybersecurity mechanisms. The data gathering was primarily conducted based on research on Scopus and Web of Science databases joint with leveraging AI for research through the ResearchRabbit platform. The most searches were directed to the keywords related to the combinations of „AI“, „cybersecurity/IT security“, „benefits“. Due to the fact that AI is very progressive topic, the searches were limited to the publications from last 5 years. By integrating findings from both academic and industry sources, this research offers a comprehensive view of the security challenges tied to AI integration in cybersecurity. The gathered data analysis is conducted through a structured thematic approach, designed to uncover recurring themes and insights from the literature. 3. Research This section is split into two subsequent parts where the first one introduces the particular use cases (areas of usage) where AI is proclaimed that it brings benefits to cybersecurity measures, whereas the second one critically reviews the real benefit of AI solutions for cybersecurity practise. 3.1 Security Areas with Potential for AI The integration of AI into cybersecurity practices has markedly transformed the landscape of organizational defense mechanisms, promising enhanced efficiency and innovation. This section explores the primary usage of AI for the cybersecurity purposes. However, the real impact of AI for cybersecurity does not always meet the benefits proclaimed by technology vendors. Below is an overview of the proclaimed security benefits connected with the deployment of AI-driven cybersecurity solutions within organizations. Simultaneously, the author also analyzed the benefits from more critical perspective, looking also on the weak points, security risks and overpromising solutions that are impressive when during their promotion, however, the reality looks different. Use case 1 – Enhanced Threat Detection and Response: AI boosts the accuracy and speed of threat detection by analyzing vast amounts of data in real-time, identifying patterns, and spotting anomalies that might signal cyber threats. It can process and correlate data from various sources, like network traffic, user behavior, and system logs, to uncover sophisticated threats that traditional methods might miss. This capability helps organizations reduce the risk of data breaches and cyber attacks, protecting sensitive information and maintaining customer trust. Faster threat detection leads to quicker response times, minimizing potential damage and downtime. Unlike traditional security systems that rely on predefined rules and signatures, AI can learn and adapt to new patterns, making it more effective in identifying unknown threats and reducing false positives. (Dhanushkodi & Thejas, 2024; Sharma, 2024; Thapaliya & Bokani, 2024) One special area is detection network anomalies. By establishing baseline patterns of normal network behavior, AI systems can identify deviations like unusual data transfers, unauthorized access attempts, or abnormal traffic patterns. Early detection of network anomalies prevents potential attacks and minimizes the impact on business operations, maintaining network integrity and ensuring the availability of critical services. Non-AI systems may rely on static rules and thresholds, which can be ineffective in detecting dynamic and evolving threats. (Alin et al., 2024; Wang et al., 2022) 66 Use case 2 –Automated Incident Response: AI automates the incident response process, cutting down the time needed to identify and mitigate threats. It can analyze security events, prioritize incidents based on severity, and initiate predefined response actions, like isolating affected systems or blocking malicious traffic. This automation reduces the workload on security teams, allowing them to focus on more complex tasks and strategic initiatives. Faster incident response minimizes the impact of cyber attacks, reducing recovery costs and maintaining business continuity. Manual incident response can be time-consuming and prone to human error, but AI-driven automation ensures consistent and rapid responses, improving overall efficiency and effectiveness. While conventional systems already offer certain automation, AI significantly enhances this by leveraging machine learning to detect new threats, adapt to evolving attack patterns, and make real-time decisions that are based on contextual analysis. Unlike rule-based systems, AI can learn from past incidents and continuously improve its response mechanisms. (Ismail, 2024; Bisht, 2024) Use case 3 – Predictive Analytics: AI uses predictive modeling techniques to anticipate potential cyber threats based on historical data and patterns. By analyzing past incidents and identifying trends, AI can forecast future threats and recommend proactive measures to strengthen defenses. This helps organizations stay ahead of cyber threats by implementing preventive measures before attacks occur, reducing the likelihood of successful attacks and enhancing overall security posture. Traditional methods often react to threats after they occur, leaving organizations vulnerable to new attacks. AI's predictive capabilities enable a forward-looking approach, improving readiness and resilience, not only by integrating real-time threat intelligence, allowing for dynamic risk prioritization and faster decision-making. This helps security teams stay focused on the most critical vulnerabilities, optimize resource allocation, and improve incident readiness. (Samia et al., 2024; Madala et al., 2023) Use case 4 – User Behavioral Analysis: AI algorithms analyze user behavior and network activity to identify deviations from normal patterns, helping detect insider threats and advanced persistent threats (APTs) that might bypass traditional security systems. AI can continuously monitor and learn from user interactions, adapting its detection mechanisms over time. AI enhances threat detection by correlating behavioral anomalies with contextual data – such as time of access, device type, or location – enabling more accurate risk assessments. It can also reduce false positives by distinguishing between legitimate unusual behavior and actual threats, improving alert quality and reducing alert fatigue among analysts. Over time, AI builds a dynamic behavioral baseline for each user, allowing for personalized security monitoring that evolves with organizational changes. Early detection of insider threats and APTs prevents significant damage and data loss, safeguarding intellectual property and sensitive information, and maintaining operational integrity and compliance with regulations. Non-AI systems may struggle to detect behavioral anomalies, especially in large and complex networks. AI's ability to learn and adapt to user behavior makes it more effective in identifying suspicious activities. (Olabanji et al., 2024; Nasir et al., 2021) Use case 5 – Vulnerability Management: AI helps identify and prioritize vulnerabilities in systems and networks by scanning and analyzing vast amounts of data. It can assess the severity of vulnerabilities and recommend remediation actions, helping organizations allocate resources effectively. Efficient vulnerability management reduces the risk of exploitation and ensures timely patching of critical vulnerabilities, maintaining a robust security posture and avoiding costly breaches. Traditional vulnerability management relies on manual processes and predefined rules, which can be slow and inefficient. AI's ability to analyze large 67 datasets and prioritize vulnerabilities enhances accuracy and speed. (Nath, 2024; Komaragiri & Edward, 2022) Use case 6 – Application Security: AI significantly enhances application security by continuously monitoring and analyzing application behavior to detect and prevent vulnerabilities and malicious activities. AI systems can identify unusual patterns and anomalies in application usage, such as unauthorized access attempts, data exfiltration, or code injection attacks. By leveraging machine learning algorithms, AI can predict potential security threats and recommend proactive measures to mitigate risks. This capability helps organizations protect their applications from exploitation, ensuring the integrity and confidentiality of data. AI-driven application security reduces the likelihood of successful attacks, minimizes downtime, and maintains user trust. Traditional application security methods often rely on static rules and manual monitoring, which can be insufficient in detecting dynamic and evolving threats. AI's ability to learn and adapt to new attack vectors makes it more effective in safeguarding applications against sophisticated threats. (Bhimanapati et al., 2024; Zhang et al., 2021) Use case 7 – Compliance & Information Security Management (ISM): AI plays a crucial role in enhancing compliance and ISM by automating the monitoring and enforcement of security policies and regulatory requirements. AI systems can analyze vast amounts of data to improve compliance with industry standards, such as GDPR, HIPAA, and ISO 27001. By continuously monitoring security controls and identifying deviations from compliance requirements, AI helps organizations maintain a robust security posture and avoid costly penalties. AI-driven ISM improves the accuracy and efficiency of compliance audits, reducing the burden on security teams and ensuring timely remediation of vulnerabilities. Traditional compliance and ISM methods often involve manual processes and periodic audits, which can be time-consuming and prone to human error. AI's ability to automate and streamline compliance monitoring enhances overall security management, ensuring organizations stay compliant with regulatory standards and protect sensitive information. (Alevizos & Ta, 2024; Reddy, 2024) 3.2 Critical Review of Real Value Proposition for Business While artificial intelligence has been researched since the mid-20th century, the recent hype surrounding AI has led to its usage as a buzzword, often overshadowing its real value proposition for businesses (Wojnar & Zbořil, 2024). Even though AI has the potential to enhance cybersecurity measures, it is often marketed as a silver bullet solution, creating unrealistic expectations (Taddeo et al., 2019). AI solutions are not plug-and-play technologies. They often require significant customization and additional training to adapt to the specific environment of an organization. Without this adaptation, AI tools may underperform or fail to address the organization’s actual security needs. Furthermore, the hype surrounding AI in cybersecurity often leads to unreasonable expectations. Marketing campaigns and media coverage tend to exaggerate AI's capabilities, presenting it as a cureall for cybersecurity challenges. This can lead to disappointment when AI systems fail to deliver on these promises. AI systems require high-quality data to function effectively. In cybersecurity, data can be noisy, incomplete, or biased, affecting the accuracy of AI predictions. AI algorithms can inherit biases present in training data, leading to discriminatory outcomes. Implementing AI solutions in cybersecurity requires significant expertise and resources, which may not be available to all organizations (Mmaduekwe, 2024; Mohammed, 2024). An example is the Use Case 7 - Compliance & Information Security Management (ISM) – where the people who create the AI solution might not 68 have experience in ISM and thus, the solution might not be trained for the traditional burdens in development of information security system (ISMS) / security governance in organizations. Moreover, over-reliance on AI can introduce new vulnerabilities. Organizations may depend heavily on AI for cybersecurity, assuming it will provide comprehensive protection. However, AI lacks human intuition, business context, and ethical awareness. It can create blind spots in defenses, as automated systems may fail to adapt to nuanced threats or understand intent (Taddeo et al., 2019). This over-reliance can undermine security efforts, as AI systems are only as good as the data they learn from and the algorithms they use. When AI is treated as unmistakable, it can lead to complacency and reduced human readiness, ultimately weakening cybersecurity measures. (Zbořil, 2024) While AI holds promise for enhancing cybersecurity measures, it is not a panacea. Businesses must approach AI with realistic expectations and understand its limitations. A balanced approach that combines human expertise with AI tools is essential for effective cybersecurity. Over-reliance on AI can lead to negative feedback, as organizations may depend on it without achieving the promised value, ultimately compromising their security posture (Enholm, 2021). 4. Conclusion The idea that companies can simply activate AI and trust it to automatically protect their systems from any threat is overly simplistic and potentially hazardous. This approach can foster a false sense of security and lead to the neglect of other vital components of a robust security strategy. The truth is, AI isn't perfect – it can be tricked, exploited, or bypassed by clever attackers. If companies depend too much on AI and ignore basic security principles and the human element, they are opening themselves up to new cyber threats. Plus, using AI without carefully considering the risks and ethical implications can lead to problems like privacy breaches or biased algorithms. Organizations need to be cautious, integrating AI into a broader security strategy rather than replacing it. AI systems further require additional training and fine-tuning to adapt to the specific use-cases and context in which they are deployed. This adaptation process can be technically complex, resourcedemanding, and may require skilled specialists with expertise in both cybersecurity and AI. These factors can significantly increase the cost and complexity of implementation, which organizations must take into consideration when planning AI integration. It's important for organizations to avoid falling for marketing hype and really think about what AI can and can't do. Before embracing AI, they should evaluate the specific issues AI can address, its limitations, and how to effectively incorporate it into a holistic security framework that includes technological, procedural, and human elements. In the end, while artificial intelligence can truly transform cybersecurity, organizations should be careful not to get caught up in promises that cannot be delivered in real. As follow up on this paper, the author will use the results as part of the long-term research aimed to materialization of cybersecurity value for business. Moreover, a potential extension of this research lies in analyzing the gap between the marketing claims of AI-powered security technologies and their actual performance and usage. References Alevizos, L., & Ta, V. (2024). Automated Cybersecurity Compliance and Threat Response Using AI, Blockchain & Smart Contracts. ArXiv, abs/2409.08390. https://doi.org/10.1007/s41870-024-02324-9. 69 Alin, G., Konsbayev, A., & Abdikaparov, N. (2024). Harnessing Artificial Intelligence for Advanced Threat Detection in Network Infrastructures. International Journal of Information and Communication Technologies. https://doi.org/10.54309/ijict.2024.18.2.007.. Bhimanapati, V., Jain, S., & Pandian, P. (2024). Security Testing for Mobile Applications Using AI and ML Algorithms. Journal of Quantum Science and Technology. https://doi.org/10.36676/jqst.v1.i2.15. Content Engine LLC. (2023). 'Catastrophic risks': At AI security summit, experts issue this warning. Bisht, R. (2024). Cybersecurity and Artificial Intelligence: How AI is Being Used in Cybersecurity to Improve Detection and Response to Cyber Threats. Tuijin Jishu/Journal of Propulsion Technology. https://doi.org/10.52783/tjjpt.v45.i02.7577. Dhanushkodi, K., & Thejas, S. (2024). AI Enabled Threat Detection: Leveraging Artificial Intelligence for Advanced Security and Cyber Threat Mitigation. IEEE Access, 12, 173127-173136. https://doi.org/10.1109/ACCESS.2024.3493957. Enholm, I., Papagiannidis, E., Mikalef, P., & Krogstie, J. (2021). Artificial Intelligence and Business Value: a Literature Review. Information Systems Frontiers, 24, 1709-1734. https://doi.org/10.1007/s10796-021-10186-w. ENISA. (2023). Cybersecurity of AI and Standardisation. Athens. Available at: [ENISA](https://www.enisa.europa.eu/publications/cybersecurity-of-ai-and-standardisation). Hannecke, M. (2023). AI Security Frameworks and Guidelines. Available at: [Medium](https://medium.com/bluetupleai/outsmarting-the-smart-guys-a-list-of-ai-security-frameworks-and-guidelines-5d126569a2fa). Ismail, W. (2024). Threat Detection and Response Using AI and NLP in Cybersecurity. J. Internet Serv. Inf. Secur., 14, 195-205. https://doi.org/10.58346/jisis.2024.i1.013. Jia, K., & Zhang, N. (2022). Categorization and eccentricity of AI risks: a comparative study of the global AI guidelines. Electronic Markets, 32(1), 59-71. Kalodanis, K., Rizomiliotis, P., & Anagnostopoulos, D. (2023). European Artificial Intelligence Act: an AI security approach. Information and Computer Security. Kaur, R., Gabrijelcic, D., & Klobučar, T. (2023). Artificial intelligence for cybersecurity: Literature review and future research directions. Inf. Fusion, 97, 101804. https://doi.org/10.1016/j.inffus.2023.101804. Komaragiri, V., & Edward, A. (2022). AI-Driven Vulnerability Management and Automated Threat Mitigation. International Journal of Scientific Research and Management (IJSRM). https://doi.org/10.18535/ijsrm/v10i10.ec05. Madala, R., Vijayakumar, N., N, N., Verma, S., Chandvekar, S., & Singh, D. (2023). Automated AI Research On Cyber Attack Prediction And Security Design. 2023 6th International Conference on Contemporary Computing and Informatics (IC3I), 6, 1391-1395. https://doi.org/10.1109/IC3I59117.2023.10397798. Mirsky, Y., et al. (2023). The threat of offensive AI to organizations. Computers and Security, 124. Mmaduekwe, U. (2024). Bias and Fairness Issues in Artificial Intelligence-driven Cybersecurity. Current Journal of Applied Science and Technology. https://doi.org/10.9734/cjast/2024/v43i64391. Mohammed, B. (2024). The impact of Artificial Intelligence on cyberspace security and market dynamics. Brazilian Journal of Technology. https://doi.org/10.38152/bjtv7n4-019. Nasir, R., Afzal, M., Latif, R., & Iqbal, W. (2021). Behavioral Based Insider Threat Detection Using Deep Learning. IEEE Access, PP, 1-1. https://doi.org/10.1109/ACCESS.2021.3118297. Nath, P., Mushahary, J., Roy, U., Brahma, M., & Singh, P. (2023). AI and Blockchain-based source code vulnerability detection and prevention system for multiparty software development. Comput. Electr. Eng., 106, 108607. https://doi.org/10.1016/j.compeleceng.2023.108607. Nature. (2023). It's time to talk about the known risks of AI. London, 618(7967), 885-886. Nigmatov, A., & Pradeep, A. (2023). The Impact of AI on Business: Opportunities, Risks, and Challenges. 2023 13th International Conference on Advanced Computer Information Technologies (ACIT), 618-622. https://doi.org/10.1109/ACIT58437.2023.10275510. Nigro, P. (2023). AI Security Risks: Separating Hype From Reality. ISBN 0890-8826. 70 Olabanji, S., Marquis, Y., Adigwe, C., Ajayi, S., Oladoyinbo, T., & Olaniyi, O. (2024). AI-Driven Cloud Security: Examining the Impact of User Behavior Analysis on Threat Detection. Asian Journal of Research in Computer Science. https://doi.org/10.9734/ajrcos/2024/v17i3424. Raval, K. J., et al. (2024). A survey on safeguarding critical infrastructures: Attacks, AI security, and future directions. International Journal of Critical Infrastructure Protection, 44, 100647. Reddy, H. (2024). Role of AI in Security Compliance. Journal of Artificial Intelligence & Cloud Computing. https://doi.org/10.47363/jaicc/2024(3)e194. Samia, N., Saha, S., & Haque, A. (2024). Predicting and mitigating cyber threats through data mining and machine learning. Computer Communications. https://doi.org/10.1016/j.comcom.2024.107949. Sharma, S. (2024). AI-Enhanced Cyber Threat Detection and Response Systems. Shodh Sagar Journal of Artificial Intelligence and Machine Learning. https://doi.org/10.36676/ssjaiml.v1.i2.14. Taddeo, M., McCutcheon, T., & Floridi, L. (2019). Trusting artificial intelligence in cybersecurity is a double-edged sword. Nature Machine Intelligence, 1, 557 - 560. https://doi.org/10.1038/s42256-019-0109-1. Thapaliya, S., & Bokani, A. (2024). Leveraging artificial intelligence for enhanced cybersecurity: insights and innovations. SADGAMAYA. https://doi.org/10.3126/sadgamaya.v1i1.66888. Wang, B., Chen, J., & Yu, C. (2022). An AI-powered Network Threat Detection System. IEEE Access, PP, 1-1. https://doi.org/10.1109/ACCESS.2022.3175886. Zaman, S., et al. (2021). Security Threats and Artificial Intelligence Based Countermeasures for Internet of Things Networks: A Comprehensive Survey. IEEE Access, 9, 94668-94690. Zhang, Z., Ning, H., Shi, F., Farha, F., Xu, Y., Xu, J., Zhang, F., & Choo, K. (2021). Artificial intelligence in cyber security: research advances, challenges, and opportunities. Artificial Intelligence Review, 55, 1029 - 1053. https://doi.org/10.1007/s10462-021-09976-0. Zbořil, M. (2024). Security risks associated with deployment of AI solutions into organizations. IDIMT: 32nd Interdisciplinary Information Management Talks. Sept. 4–6, 2024. Hradec Králové, Czech Republic. DOI: 10.35011/IDIMT-2024-65 Wojnar, M. & Zbořil, M. (2024). How to not fall for marketing illusions in the world of cybersecurity? Information Security Summit (IS2). May 29-30, 2024. Prague, Czech Republic. 71 APPLYING PROTECTION MOTIVATION THEORY TO ENHANCE ADHERENCE TO ONLINE SECURITY PRACTICES Tomáš Sigmund Prague University of Economics and Business Faculty of Informatics and Statistics [email protected] DOI: 10.35011/IDIMT-2025-71 Keywords Protection motivation theory; online security; threat appraisal; coping appraisal Abstract With the increasing frequency and sophistication of cyber threats, individual compliance with online security practices has become a critical factor in organizational and societal cybersecurity. This article explores the application of Protection Motivation Theory (PMT) as a theoretical framework for understanding and influencing user behavior in the context of online security. We examine the core components of PMT—threat appraisal and coping appraisal—and their predictive power in shaping security-related intentions and behaviors. By synthesizing findings from empirical studies and incorporating original survey-based research, we highlight determinants of user motivation to engage in following online security measures. We also explore how these findings can be translated into actionable strategies. 1. Introduction As digital technologies permeate nearly all aspects of life, individuals face growing exposure to online security threats. Despite the availability of advanced technical solutions—such as intrusion detection systems, firewalls, automation, artificial intelligence and encryption protocols—human behavior continues to represent a significant point of vulnerability (Moustafa et al., 2021), (Kim & Kim, 2024). Even well-informed users sometimes bypass security mechanisms for the sake of convenience, creating an ongoing challenge for security professionals and policymakers. Research in behavioral cybersecurity has emphasized the need to account for psychological, emotional, and social variables that influence individuals’ compliance with recommended practices. In this context, Protection Motivation Theory (PMT), developed by Rogers (1975) and revised in 1983 (Rogers et al., 1983) to explain individual health behavior, offers a compelling framework for understanding how users evaluate cyber threats and determine whether to engage in protective actions. The theory’s emphasis on cognitive appraisal processes aligns well with the decision-making challenges posed by cybersecurity behavior, making PMT a relevant model in the digital age. Its adaptability across diverse domains of behavior—including health, environment, and now cybersecurity—demonstrates its conceptual strength. Moreover, PMT allows for nuanced 72 understanding of motivational and deterrent factors, providing a diagnostic lens for designing personalized security interventions. While this article includes several recommendations for improving online security behavior based on PMT, it is important to clarify that not all of them emerge directly from our survey data. Some proposed interventions are supported by the correlations identified in our findings, particularly those related to coping appraisal constructs. Others are theoretically grounded in PMT or derived from related literature in behavioral science and information security. 2. Theoretical Background: Protection Motivation Theory Protection Motivation Theory posits that individuals are motivated to protect themselves from harm when they cognitively appraise both the nature of the threat and their ability to cope with it. These appraisals are broken down into two core dimensions: Threat Appraisal, which assesses the danger posed by a cyber threat and includes: • Perceived severity: an individual's judgment regarding the seriousness of the consequences if the threat materializes. • Perceived vulnerability: an individual’s belief about their likelihood of being personally affected. • Rewards of non-compliance: benefits associated with not taking protective action, such as convenience or saved time. Coping Appraisal, which evaluates one’s ability to deal with the threat, consisting of: • Response efficacy: the perceived effectiveness of the recommended protective behavior. • Self-efficacy: the individual’s confidence in their ability to carry out the behavior. • Response costs: the perceived barriers or drawbacks to engaging in the behavior, such as time, effort, or financial resources. Protection Motivation Theory (PMT) suggests that individuals are more likely to engage in protective behaviors when they perceive a threat as both severe and personally relevant, when they do not expect to gain any rewards from non-compliance, and when they believe that taking protective action is both effective and affordable. 3. PMT and Online Security Compliance Over the past two decades, PMT has been applied in a range of contexts including health communication (Floyd et al., 2000), environmental behavior (McCaughey et al., 2017), and increasingly, information security. Digitalisation and advancements of technologies triggered the wide employment of the protection motivation theory in the information systems management field over the last two decades. Considering the debates about the threats that technology could pose for users (e.g. pretexting, phishing, targeted malware, cyberattacks) and organisations (Verkijika, 2018), (Ifinedo, 2012), (Thompson et al., 2017), the theory was useful in understanding the factors that make individuals avoid technology-related threats (Crossler, 2010). Numerous empirical studies support the theory's utility in explaining user behavior in online environments. In the realm of cybersecurity, PMT has been used to investigate behaviors such as password hygiene, email safety, and software maintenance. The modular nature of PMT also allows integration with other behavior change theories ETHICAL ASPECTS OF AI: RESEARCH AND USAGE 81 SELECTED ETHICAL ASPECTS OF AUTHORSHIP OF THE USE OF AI PRODUCTS Anton Lisnik Slovak University of Technology in Bratislava Institute of Management [email protected] DOI: 10.35011/IDIMT-2025-81 Keywords Authorship, AI products, Ethics. Abstract The use of AI brings with it many conveniences and advantages for every AI user. However, information processing is associated with the quantity and quality of resources and the ways in which artificial intelligence learns. Therefore, there are many new questions and challenges in scientific work. Among them are text authorship and bias. The article deals with these two phenomena in connection with scientific practice. 1. Introduction. The development of the use of AI in many areas has become an integral part of scientific work and scientific research. In this article, I will discuss the two most debatable threats to the use of AI in science, which are bias and privacy protection. AI is most often used in the creation of science to process large volumes of data, perform simulations and predictions that would be unfeasible for traditional research methods. With the development of artificial intelligence (AI), new horizons are opening up in the field of science, from medicine to astrophysics to biotechnology. AI technologies allow scientists to process huge amounts of data, perform simulations and predictions that would be unfeasible for traditional research methods. However, with this technological revolution also come significant ethical challenges that deserve attention to ensure that the use of AI in science is not only effective, but also responsible. (Katrencik, 2023) 2. Selected references for ethical principles for the use of artificial intelligence in scientific research The use of artificial intelligence in scientific research should be guided by broad ethical principles and guidelines. Various international and national frameworks provide an important basis for the responsible and ethical use of AI. The UNESCO Recommendation on the Ethics of Artificial Intelligence (UNESCO, 2021) presents ten core principles that emphasize a human rights-based 82 approach to the ethics of AI. These principles include proportionality and non-harmfulness, safety and security, the right to privacy and data protection, multi-stakeholder and adaptive governance and collaboration, responsibility and accountability, transparency and explainability, human oversight and determination, sustainability, information and literacy, and equity and non-discrimination (UNESCO, 2021). Together, these principles form the basis for the ethical use of AI systems, always keeping in mind the importance of human oversight. (UNESCO, 2021)(Fjelland, 2020) In addition, the Belmont Report (IBM, 2024) (Giroux, 2022), originally developed for human research, is proposed as a historical precedent for ethical research in AI. The three core principles of the Belmont Report – respect for persons, beneficence and justice – could be directly applied to AI research involving human subjects, thereby promoting more trustworthy and responsible uses of AI. Respect for persons would require informed consent from individuals whose data is used to train AI systems. Beneficence would mean designing studies to minimize risks to participants, and justice would require fair selection of subjects, avoiding inappropriate exclusion that can lead to bias in the data. (IBM, 2024)(Giroux, 2022) IBM has set out three guiding principles for the development of data and AI: the purpose of AI is to augment human intelligence, data and knowledge belong to their creators, and AI systems must be transparent and explainable (IBM, 2024). These principles emphasize collaboration between humans and AI, ownership of data, and the importance of transparency in AI recommendations (IBM, 2024). Similarly, the Principles for the Ethical Use of AI in the United Nations System (CEB, 2024) are based on UNESCO recommendations and highlight ten principles rooted in ethics and human rights. 2.1. Bias One of the biggest risks when using AI in scientific research is the potential presence of bias. Algorithms are built on historical data, and if that data contains biases (such as discrimination against certain groups or unfair representations), AI can carry them into its decisions. This can be particularly dangerous in medicine, where incorrect or unfair decisions can lead to serious consequences for individuals or even entire populations. An example would be an algorithm that helps diagnose diseases but has been trained on data that only includes certain ethnic groups, which may lead to lower diagnostic accuracy in patients from other ethnic groups. The solution to this challenge is to ensure that algorithms are trained on diverse and representative data and that they are regularly monitored and updated. Bias can seriously affect the veracity of information in relation to investing. (Katrenčík, 2018) Three categories of bias in computer systems have been developed from real-world case studies: preexisting, technical, and emergent. Pre-existing bias has its roots in social institutions, practices, and attitudes. Technical bias arises from technical constraints on reasoning. Emergent bias arises in the context of use. Although other authors have pointed to bias in specific computer systems and have pointed to a general problem, one solution is to examine the extent of bias and consider how to deal with the presence of bias, and it should be included among a selected set of criteria—including reliability, accuracy, and efficiency—by which the quality of systems used in society should be judged. (Friedman, 1996) “Respect for persons”, which includes autonomy, refers to the right of an individual to make decisions regarding their physical body and its derivatives (e.g., tissue samples) and their personal information. Autonomy is usually respected through the practice of informed consent. In the context of artificial intelligence, autonomy must be taken into account when using personal health data for the development, training and validation of artificial intelligence systems and when applying artificial intelligence systems in patient care. Many individuals expect transparency and control over their 83 identified and even anonymised data. In the European Union, the enforcement of means of arbitrary deletion has been demonstrated through their “right to erasure (to be forgotten)”. This could help balance people’s autonomy and the development of artificial intelligence. (Rocher, 2019) Furthermore, as artificial intelligence technologies become increasingly widespread in many areas, concerns may also arise regarding people’s awareness of and consent to AI-assisted diagnostics. Type bias: AI bias can manifest in various ways, affecting different demographic groups and societal sectors. Some common types of AI bias include: 1. Data Bias: This occurs when the data used to train AI models is not representative of the real world or reflects existing societal biases. For example, if a hiring algorithm is trained on historical data where men were predominantly in leadership roles, it might unfairly favour male candidates. Different types of data bias include: a) Historical bias: Arises from past prejudices reflected in the data. b) Sample bias: Occurs when the training data doesn't accurately represent the population the AI will interact with. c) Label bias: Happens when data labelling is inconsistent or reflects subjective opinions. d) Measurement bias: Occurs due to inaccuracies or incompleteness in data collection and recording. e) Exclusion bias: Arises when certain groups or information are left out of the training data. f) Selection bias: Occurs when the data collection process favours certain groups. g) Coverage bias: A type of selection bias where some groups are underrepresented in the dataset. h) Non-response bias: Another form of selection bias where certain individuals or groups are less likely to participate in data collection. 2. Algorithmic Bias: Even with unbiased data, the design of algorithms can introduce bias. This can happen through the choice of features, how different variables are weighted, or the inherent assumptions in the algorithm's structure. 3. Human Bias: The biases and preconceived notions of the people involved in creating, training, and deploying AI systems can unintentionally seep into the technology. This can occur during data labelling, feature selection, or even in how the problem itself is framed. Different types of human bias that can affect AI include: a) Prejudice bias: When societal stereotypes and faulty assumptions are included in the data. b) Recall bias: Occurs during data labelling when labels are applied inconsistently due to subjective observations. c) Confirmation bias: When developers or the AI system itself favours information that confirms pre-existing beliefs. d) Automation bias: The tendency to over-rely on AI systems and disregard contradictory information. e) In-group bias: Favouring individuals or data from groups one identifies with. f) Out-group homogeneity bias: Perceiving members of other groups as more similar than they actually are. 4. Representation Bias: This occurs when certain groups are underrepresented or overrepresented in the training data, leading to skewed results for those groups. 84 5. Interaction Bias: Arises from the way users interact with the AI system, where different groups might use the system in ways that lead to biased outcomes. Causes of AI bias are multifaceted and often interconnected: • Biased Training Data: As mentioned, if the data used to train AI reflects existing societal biases or doesn't represent the diversity of the real world, the AI will likely perpetuate and even amplify these biases. This can stem from historical inequalities, skewed representation, or flawed data collection processes. • Flawed Algorithm Design: The way an algorithm is designed, including the features it prioritizes and how it weighs different factors, can inadvertently introduce bias, even with seemingly neutral data. • Human Oversight (or Lack Thereof): While human review can help catch biases, if the humans involved hold their own biases (conscious or unconscious), these can still influence the AI system. Additionally, a lack of sufficient human oversight can allow biases to go unnoticed and uncorrected. • Lack of Diversity in Development Teams: When AI systems are developed by teams lacking diverse perspectives, unconscious biases are more likely to be embedded in the technology. Mitigating AI bias is a complex but crucial endeavour. Some strategies include: • Ensuring Diverse and Representative Data: Carefully curating training datasets to accurately reflect the population the AI will serve is essential. This includes actively seeking out and incorporating data from underrepresented groups. • Algorithmic Auditing and Transparency: Regularly testing and evaluating AI systems for bias is crucial. Explainable AI (XAI) techniques can help understand how AI models make decisions, making it easier to identify potential sources of bias. • Diverse Development Teams: Building AI systems with diverse teams can bring a wider range of perspectives to the development process, helping to identify and mitigate potential biases. • Feature Engineering and Selection: Carefully selecting and engineering the features used by the AI model can help avoid proxies for sensitive attributes that might lead to bias. • Bias Detection and Correction Techniques: Various technical methods are being developed to detect and mitigate bias in AI models during training and deployment. • Establishing Ethical Guidelines and Governance: Organizations should prioritize the ethical and responsible use of AI and establish clear guidelines and oversight mechanisms to address bias. • Human-in-the-Loop Systems: Incorporating human review and feedback in the AI development and deployment process can help identify and correct biases. (Rocher, 2019). 2.2 Privacy and data security All data that is part of the content of the virtual space is always associated with the personalization of the creator of the information, or the person making the information available to the viral space. In fact, every activity is associated with a trace that can be used to trace the author and also trace all data 85 about the author and also about the place, time of creation of the publication and identification of the device, access point from where these changes to the information occurred. A special ethical challenge is also the use of AI in scientific research, especially in areas such as genetics or personalized medicine, which often requires access to sensitive personal data. It is essential to ensure the protection of this data to prevent its misuse or leakage. Researchers must adhere to strict ethical and legal standards to protect the privacy of research participants and at the same time ensure that their research does not lead to the creation of algorithms that could encounter privacy problems, such as unauthorized tracking or profiling of individuals. The phenomenon of anonymity and the investigation of authorship The phenomenon of anonymity is, as the Internet has grown, so have questions of privacy and authorship. Ensuring anonymity and authorship allows individual Internet users to carry out their online activities in comfort and privacy (Kambourakis, 2014). According to Burkell (Anonymity in Behavioural Research: Not Being Unnamed, But Being Unknown, 2006), anonymity is at one level synonymous with the impossibility of identifying an individual. After all, the word anonymous itself means “without a name”. However, if we interpret the term in a slightly broader sense, we realize that it also includes a meaning reflecting social distance and at the same time a lack of distinguishableness from those around us. In the first sense, anonymity contrasts with the condition of being named, and in the second with the state of being known (Burkell, 2006). When considering how anonymity affects user behavior on the Internet, according to Burkell (2006), it is useful to look at it from three main perspectives: 1. Name anonymity, defined as the non-disclosure of identifying information, has relatively clear behavioral consequences: people are willing to provide sensitive information and are more likely to be honest (Burkell, 2006). 2. Visual anonymity, mainly related to facial unrecognition, can lead to a reduced perception of interpersonal obligations, a greater willingness to reveal oneself, higher levels of aggression and a lower willingness to help in interpersonal interactions (Burkell, 2006). 3. Anonymity of actions consists in the fact that an individual’s actions are either not seen or cannot be attributed to him. The consequence of this anonymity is a great sense of freedom and independence associated with the lowest possible level of responsibility (Burkell, 2006). In some areas, information about the authorship (ownership) of information is even required. For example, when we are looking for information about the originator or owner of a thing or statement. (Compare Zatrochová, 2021) In the scientific field, anonymity is an unwanted phenomenon. Authors want to be identified with their ideas by authorship, because it is their property. The use of AI is problematic in relation to the anonymization of the author. The way in which AI creates content information, despite the method of traceability, it is impossible to fully identify authorship, and so what was supposed to serve the development becomes a direct threat to the values for which online repositories and platforms for sharing information, knowledge and processes in science were created. There are still areas in which we want to maintain anonymity, it is even required, because it concerns personal information such as health, social status, origin and even opinions and ideas. Building ethics is necessary from the beginning of education. (Osvaldová, 2022) 3. Methodology In the research, we selected the most commonly perceived threats to the use of AI in scientific practice. We identified the most common threats: autonomy, bias, transparency, privacy protection 86 and employment outflow. We evaluated selected aspects among academics in Slovakia at the Slovak University of Technology in Bratislava. We conducted the research on a sample of 100 academic respondents in various academic positions and in various years of study. The main distribution of the sample was 50 academics and 50 students. The research was conducted in the months of December 2024 - January 2025. A maximalist scale of 1-5 was chosen. As the first research task, we dealt with the arrangement of threats. In the research, I will deal with the most perceived threats and these are the ones identified in the first places from the research: bias and text authorship. We investigated how respondents perceive the importance of truth or error in statements found using AI and how they perceive the principles of authorship in content creation. 4. Research We have established the following hypothesis: H0: Respondents in the monitored sets have comparable perceptions of bias. H1: Respondents in the monitored sets do not have comparable perceptions of bias. To test the hypotheses, the Anova test was used at a significance level of 0.05. The test results are presented in Table 1. Table 1 - ANOVA program output Groups Count Sum Average Variance Students 50 276 3,65 0,7875 Academics 50 290 3,83 0,7050 Source of Variation SS df MS F P-value F crit Between Groups 23,89776 8 2,98722 4,081064 9,55E05 1,954029 Within Groups 433,3269 592 0,731971 H1 Total 457,2246 600 Source: own processing Based on the values of F and Fcrit, we recommend accepting the alternative hypothesis H1. H1: Respondents in the monitored sets do not have comparable perceptions of bias. Student respondents show significant differences in perceptions of bias threat. Using the correlation coefficient, they examined the degree of agreement between perceptions of bias threat and authorship (anonymity). 87 Table 2Corelations Score Q1 Average Q r Students 75,2 67,64706 0,30443 Academics 75,2 68,72941 0,70058 Source: own processing The correlation shows that academics are more concerned about the loss of the relationship between information and authorship than students. 5. Conclusion We can assume that students do not place as much emphasis on attribution and do not fear the loss of authorship to the same extent as academics. This is a natural consequence of the importance of consequences. We can therefore assume that these things arose on the basis that the threat to the existence of students in relation to the results of their work is not as great as for academics. The use of AI in science offers great potential for accelerating discoveries and progress in various fields. However, like any new technology, it also brings with it a number of ethical challenges. It is essential that scientists, developers and regulators work together to create frameworks that ensure the responsible and ethical use of AI. Only in this way can we avoid potential negative consequences and maximize the benefits of this technology for society. Acknowledgement Supported by a grant KEGA: 023STU-4/2023 References Burkell, J. (2006). Anonymity in Behavioural Research: Not Being Unnamed, But Being Unknown. University of Ottawa Law & Technology Journal, pp. 189-203. CEB 2024. In: https://unsceb.org/principles-ethical-use-artificial-intelligence-united-nations-system Fjelland, R. Why general artificial intelligence will not be realized. Humanit Soc Sci Commun 7, 10 (2020). https://doi.org/10.1057/s41599-020-0494-4 Friedman, B., Nissenbaum, H.: Bias in computer systems. In: ACM Transactions on Information Systems (TOIS), Volume 14, Issue 3. (1996). Pages 330 – 347. https://doi.org/10.1145/230538.230561 Giroux, M., Kim, J., Lee C, J., Park, J. Artificial Intelligence and Declined Guilt: Retailing Morality Comparison Between Human and AI. In: Journal of Business Ethics (2022) 178 :1027-1041. (1). July 2022. DOI: 10.1007/s10551-02205056-7 IBM (2024) IBM principles for data and AI. In: https://www.ibm.com/think/reports/ai-in-action. Kambourakis, G. (2014). Anonymity and closely related terms in the cyberspace: An analysis by example. Journal of Information Security and Applications, pp. 1-16. Katrencik, I., Mucha, B., Zatrochova, M.: Ethical implications of artificial intelligence data usage: a case study of Slovakia and global perspetives. In: IDIMT 2023: New Challenges for ICT and Management - 31st Interdisciplinary Information Management Talks. Pages 365 - 3722023 31st Interdisciplinary Information Management Talks: New Challenges for ICT and Management, IDIMT 2023Hradec Kralove6 September 2023through 8 September 2023. ISBN: 978-399151176-2, DOI: 10.35011/IDIMT-2023-365 88 Katrenčík, I., Zatrochová, M.: Use of cryptocurrency in business. Current problems of the corporate sector 2018 [16.05.2018-17.05.2018, Bratislava, Slovensko]. In: Aktuálne problémy podnikovej sféry 1. vyd. – Bratislava (Slovensko) : Ekonomická univerzita v Bratislave. Celouniverzitné pracovisko EUBA. Vydavateľstvo EKONÓM, 2018. – ISBN 978-80-225-4536-5, s. 501-511 Osvaldová, Z., Závadský, J: Rozvoj etiky a spoločenskej zodpovednosti vo vzdelávacích organizáciách. In: Príklady dobrej praxe etiky a aplikácie normy ISO 37001: nekonferenčný zborník odborných prác 1. vyd. – Poprad (Slovensko) : Výskumný ústav ekonomiky a manažmentu, 2022. – ISBN 978-80-973663-3-9, s. 32-62 Rocher L, Hendrickx JM, de Montjoye YA. Estimating the success of reidentifications in incomplete datasets using generative models. Nat Commun. 2019;10(1):3069. https://doi.org/10.1038/s41467-019-10933-3) UNESCO. (2021). Odporúčanie UNESCO o etike umelej inteligencie. https://unesdoc.unesco.org/ark:/ 48223/pf0000380457 Zatrochová, M., Kuperová, M., Golej, J.: Analysis of the principles of reverse logistics in waste management. In Acta Logistica. Vol. 8, Iss. 2 (2021), s. 95-106. ISSN 1339-5629 (2021: 0.240 - SJR, Q3 - SJR Best Q). WOS: 000643042600001; SCOPUS: 2-s2.0-85110661575; DOI: 10.22306/al.v8i2.208. 95 Figure 1. Relationship Between Ethical Concerns and Academic AI Acceptance Source: author own research 4. Conclusion This study explored university students’ perceptions and attitudes toward artificial intelligence (AI) in education, focusing on self-perceived knowledge, usage patterns, ethical concerns, and acceptance of AI in academic environments. The findings provide valuable insights into how students across disciplines are engaging with AI and what concerns they have when this technology becomes increasingly integrated into higher education. We found that students in technical (STEM) programs reported significantly higher self-perceived knowledge of AI compared to those in non-STEM fields, although the actual frequency of use did not differ significantly between the two groups. This suggests that confidence in understanding AI can be formed more by academic background than by daily use. Students who frequently engage with AI tools were more open to its use in academic contexts, especially in tasks such as generating outlines or ideas for written work. Interestingly, their ethical concerns - particularly regarding privacy, data usage, and fairness - were not significantly different from those of infrequent users. This may indicate that familiarity with AI breeds a certain degree of trust or at least practical acceptance of its utility. Finally, we observed a clear, statistically significant negative link between ethical concern and acceptance of AI in academic evaluations. Students with stronger concerns about the social and ethical risks of AI were pariculary less likely to support its integration into formal academic work. This emphasizes the importance of ethical awareness in shaping attitudes toward AI and suggests that an open and ongoing dialogue about the risks and responsibilities associated with AI use in education is crucial. Although the results of this study provide valuable insights into students’ perceptions and ethical attitudes toward AI in higher education, some limitations must be recognised. The sample included 103 respondents, the number of master’s level students was too low to be retained in the analysis. 96 Therefore, the study focused exclusively on bachelor’s-level participants, which may limit the generalizability of the findings across different levels of academic maturity. All responses were selfreported, which is the usual limitation of social desirability bias and subjective interpretation. For example, students' self-assessed knowledge of AI may not accurately reflect their actual understanding or competence. Future research should seek to expand both the scope and depth of research into students’ engagement with AI in educational contexts. First, the inclusion of a more diverse and larger sample - including master’s and doctoral students, as well as participants from different institutions or countries - would increase the generalization. Secondly, it would be beneficial to conduct longitudinal studies that track how students’ knowledge, use patterns, and ethical perspectives on AI evolve over time, especially as institutional policies and AI tools develop. Acknowledgement The contribution is a partial output of the KEGA research task no. 023STU-4/2022 conducted at the STU in Bratislava. References Annenberg School for Communication and Journalism. (2024). Ethical dilemmas in AI. University of Southern California. https://annenberg.usc.edu/research/center-public-relations/usc-annenberg-relevance-report/ethical-dilemmas-ai Agarwal, B., Urlings, C., Lankveld, G., & Klemke, R. (2024). Identifying the ethical values and norms of artificial intelligence in education: a systematic literature review.. https://doi.org/10.35542/osf.io/e7t3f Barnes, E. and Hutson, J. (2024). Navigating the ethical terrain of ai in higher education: strategies for mitigating bias and promoting fairness. Forum Edu. Stud., 2(2), 1229. https://doi.org/10.59400/fes.v2i2.1229 Bennett, S., & Bell, T. (2024). Striking a balance: Navigating the ethical dilemmas of AI in higher education. EDUCAUSE Review. https://er.educause.edu/articles/2024/12/striking-a-balance-navigating-the-ethical-dilemmas-of-ai-inhigher-education Duah, J. and McGivern, P. (2024). How generative artificial intelligence has blurred notions of authorial identity and academic norms in higher education, necessitating clear university usage policies. International Journal of Information and Learning Technology, 41(2), 180-193. https://doi.org/10.1108/ijilt-11-2023-0213 Freeman, J. (2025). Student Generative AI Survey 2025. Higher Education Policy Institute. https://www.hepi.ac.uk/2025/02/26/student-generative-ai-survey-2025/ Holmes, W., Persson, J., Chounta, I.-A., Wasson, B., & Dimitrova, V. (2022). Artificial intelligence and education: A critical view through the lens of human rights, democracy, and the rule of law. Council of Europe. https://rm.coe.int/artificial-intelligence-and-education-a-critical-view-through-the-lens/1680a886bd Holmes, W., Iniesto, F., Anastopoulou, S., & Boticario, J. (2023). Stakeholder perspectives on the ethics of ai in distancebased higher education. The International Review of Research in Open and Distributed Learning, 24(2), 96-117. https://doi.org/10.19173/irrodl.v24i2.6089 Lisník, A., & Majerník, M. (2023). Aspects of strategic management and online marketing. In Developments in information and knowledge management systems for business applications (pp. 49–80). Springer Cham. Nidhi, N., Arenas, M., Morgado, A., Solomon, A., Moreira, M., Ferreira, R., … & Raittila, E. (2024). Game-based learning as a tool for teaching ethical ai to youth: insights from the charlie project. European Conference on Games Based Learning, 18(1), 658-667. https://doi.org/10.34190/ecgbl.18.1.2847 Perkins, M. and Roe, J. (2023). Decoding academic integrity policies: a corpus linguistics investigation of ai and other technological threats. Higher Education Policy, 37(3), 633-653. https://doi.org/10.1057/s41307-023-00323-2 Yu, L. and Yu, Z. (2023). Qualitative and quantitative analyses of artificial intelligence ethics in education using vosviewer and citnetexplorer. Frontiers in Psychology, 14. https://doi.org/10.3389/fpsyg.2023.1061778 97 ETHICAL ASPECTS OF ARTIFICIAL INTELLIGENCE AND THE DEVELOPMENT OF ENTREPRENEURSHIP AMONG YOUNG PEOPLE Krzysztof M. Krusiec Slovak University of Technology in Bratislava Institute of Management [email protected] DOI: 10.35011/IDIMT-2025-97 Keywords Ethics; artificial intelligence; entrepreneurship; ChatGPT; personal development Abstract The 21st century has brought numerous changes in the replacement of human labor with artificial intelligence-based solutions, raising serious ethical dilemmas related to mass layoffs, the elimination of long-anticipated job positions for graduates, and the dehumanization of the work process. In light of these issues, the aim of this paper is to conduct an in-depth analysis of the impact of using artificial intelligence tools - particularly generative chatbots such as ChatGPT - on the development of entrepreneurship among young people. In the face of increasing automation of cognitive and professional processes, both the potential and the risks of replacing human activity with algorithms are highlighted. Based on a review of the relevant literature and statistical data (including Eurostat and GEM), a research model was developed and then subjected to empirical verification. The study was conducted among 210 individuals aged 20 to 30 who use AI-based chatbots. A probit model was applied to identify the relationship between the intensity of AI usage and the level of cognitive independence. The results showed that excessive reliance on AI chatbots fosters passivity and a decline in initiative, whereas their creative use may contribute to the development of entrepreneurial competencies. 1. Ethical implications of artificial intelligence in the development of entrepreneurship Will the rapidly changing socio-technological reality bring about an adaptation crisis for parts of 21stcentury society? Younger generations, raised with the belief in limitless opportunities, often collide with the reality of precarity - lack of job stability, employment below qualifications, and limited prospects for independence. The issue lies not in the qualifications themselves, but in how they are assessed by employers, often based on subjective criteria and economic calculations. This raises serious ethical dilemmas, such as whether it is justified to keep a competent employee in a lowranking position for the sake of company profit (Palej, Krusiec, 2024). Overlaid onto this context is the rapid development of AI, which already performs many tasks - from data analysis to content creation and even medical diagnostics - more efficiently than humans (Barney, 2024). With the low cost of tools like ChatGPT Plus ($20-200/month), the boundary between the necessity of human input 98 and its replaceability by technology has been crossed. Generation Z, particularly attuned to the digital environment, now faces a labor market that demands completely new qualifications (Kuperová, Lisnik, 2023; World Economic Forum, 2025). A striking fact is that in 2022, 28% of European Union (EU) residents lived in poverty or were at risk of social exclusion (Eurostat, 2024), evoking the historical phenomenon of “white slavery” (Krawczyńska, 1930). From an ethical standpoint, excessive subordination of humans to the logic of profit threatens their marginalization. The World Economic Forum's 2016 statement - “I own nothing, have no privacy, and life has never been better” - evokes associations with feudalism and systemic social dependency (World Economic Forum, 2016). Although we formally possess rights, actual control over our own fate is becoming increasingly limited. What becomes essential is not only nominal income, but also the ability to manage resources, which ensures psychological comfort and stability (Krusiec, 2018). In the face of automation, the development of entrepreneurship may serve as a pathway to regaining agency - not only as a defense against crisis, but also as a means of reclaiming control over one’s life. A case in point for the need to strike a balance between AI and human labor was the Covid-19 pandemic, which necessitated the expansion of remote work systems (Khusainov, Lisnik, 2023). 2. The potential of artificial intelligence One of the key aspects of modern artificial intelligence is its intellectual advantage over a significant portion of the population. The potential of artificial intelligence becomes particularly evident in processing vast streams of data generated by IoT sensors, enabling quick responses to environmental changes and increasing the efficiency of decision-making processes (Bawa, Špirková, 2016). In an interview conducted with the ChatGPT Plus model, when asked about its own cognitive potential, it assessed its IQ at over 140 points, emotional intelligence at 110 points, language skills above 150 points, and general knowledge between 140 and 150 points (Conversation with ChatGPT Plus, 2025). This kind of self-assessment not only draws attention to the high capabilities of AI but also highlights a fundamental difference between humans and machines - namely, that the tool requires no sleep, rest, or emotional support. It is capable of uninterrupted work, regardless of stress or pressure, which often impact the quality and efficiency of tasks performed by humans (Bryant, 2023). AI no longer serves merely an auxiliary function but increasingly becomes an irreplaceable, independently “thinking” partner in daily work and life. A project that would take a person hours to complete can be carried out by artificial intelligence in just a few minutes - often with greater precision and without external intervention. This opens up entirely new perspectives in many fields, such as interior design, instant data mining from the internet, or the development of educational materials. Such support in conducting scientific research shows why ChatGPT is gradually being introduced as a teaching tool in educational institutions (Adel, Ashan, Davison, 2024). Literature also indicates that young entrepreneurs are increasingly using AI-based tools, such as data processing systems and chatbots. Therefore, ethical data management becomes essential, built on transparency, respect for privacy, and the responsible use of information (Lisnik, Kuperová, 2023). These issues are also important because cryptocurrencies are becoming more popular among investors, which entails high ethical risks (Katrenčík, Zatrochová, 2017). Increasingly, questions are also being raised about the limits of AI application in both social and private spheres, as in the case of biometric systems used on social media platforms (Lisnik, Janíčková, Zimermanová, 2020). The ethical use of such tools requires not only technical precision but, above all, awareness of the consequences - from data protection to the threat of surveillance or manipulation of user behavior, etc. (Krusiec, 2019). Is it possible to create a balance between technological progress and the protection of human dignity and rights? If AI development continues to follow solely a market logic 99 - maximizing profits while minimizing costs - this may lead to a situation in which a large part of society is deprived of income sources, control over their future, and the sense of purpose in their work. This presents a real threat of deepening economic exclusion, social inequalities, and a new form of digital alienation of the human being, in a world it fosters but gradually loses control over. 3. Entrepreneurship and artificial intelligence Entrepreneurship is not merely the ability to run a business, but a set of competencies - such as initiative, creativity, responsibility, and flexibility - that enable effective functioning in a changing world. Developing these qualities supports individuals in fulfilling social and professional roles, as well as in creating their own life space. The challenges inherent in entrepreneurial activity - stress, risk of failure, financial uncertainty - require appropriate educational support that builds resilience, innovation, and a civic mindset. One of the tools that can both support and threaten this development is artificial intelligence. AI enhances many processes: it analyzes data, generates content, supports business decisions, and reduces costs, making it an attractive option for novice entrepreneurs (GEM, 2025). However, its overuse can lead to cognitive passivity, erosion of critical thinking, and superficiality in action. Instead of developing their own intuition and sense of responsibility, users may overly rely on AI models that bear no consequences for their mistakes. This raises the question: does the development of AI truly support humans, or does it rather create dependency? A balanced model is needed - one that integrates the potential of technology with the cultivation of responsible and conscious users. This is especially important given that the benefits of AI are not distributed evenly. It is primarily global corporations - capable of implementing advanced technologies - that gain the most. Local businesses, often family-owned and with limited resources, are being pushed out of the market, leading to increasing inequality and a decline in economic diversity - contrary to the principles of social ethics. 4. Methodology The aim of this study is to conduct an in-depth analysis of the impact of using artificial intelligencebased tools particularly generative chatbots such as ChatGPT - on the development of entrepreneurship among young people. Given the rapid advancement of digital technologies and the increasing automation of cognitive and decision-making processes, fundamental questions arise regarding the extent to which modern technological solutions support - or, paradoxically, may limit - the development of entrepreneurship, independence, and creativity among individuals entering the labor market. The research is guided by the following question: does intensive use of AI-based chatbots, such as ChatGPT, support the development of entrepreneurship among young people, or - when viewed through an ethical lens - can it lead to a weakening of autonomy, creativity, and individual responsibility for one’s decisions and actions? This issue is interdisciplinary in nature, touching upon areas such as behavioral economics, cognitive psychology, and entrepreneurial education. These considerations are particularly relevant in the context of civilization challenges tied to digital transformation and the growing importance of future skills. The study involved 210 individuals aged 20 to 30 who declared regular use of AI-based chatbots, particularly ChatGPT. Respondents were selected purposively - from among students and young 100 professionals representing various industries and sectors. All participants remained anonymous, which helped minimize self-presentation bias and increased the reliability of the collected data. The research tool used was a proprietary questionnaire consisting of 21 closed and semi-open questions. The questions focused on, among other things, the frequency of AI usage, the perceived effects of this practice on work, learning, self-efficacy, as well as changes in decision-making and problem-solving strategies. Special attention was paid to potential symptoms of cognitive passivity and the extent to which algorithmic thinking was replacing independent thought. To determine the impact of AI on selected behaviors, a probit model was applied - a statistical tool for analyzing binary variables, which makes it possible to identify the relationship between independent variables (e.g., intensity of AI chatbot use) and a dependent variable (e.g., exhibiting the trait of independence: 0 = absent, 1 = present). This allowed for a precise assessment of the strength and direction of the examined relationships while maintaining scientific rigor. The research project was carried out with the substantive support of the “Instytut Rozwoju Przedsiębiorczości im. Krzysztofa M. Krusiec”, a non-governmental organization established in 2020 to supplement knowledge and competencies in the field of entrepreneurship, particularly among young people. The mission of the “Fundacja IRP” is personal and business development through a proprietary, multidisciplinary life improvement, development and stabilization system known as the “Model K. M. Krusiec” (Krusiec, Palej, 2024). As part of the conducted study, the following specific research questions were formulated: • Does the use of AI-based chatbots promote the development of entrepreneurship among individuals at risk of precarization, or does it rather lead to cognitive passivity and avoidance of independent problem-solving? • Which specific competencies and traits related to entrepreneurship (e.g., initiative, organizational ability, creativity) can be supported by regular use of artificial intelligence tools? • What potential risks and deficits may arise from frequent reliance on AI in performing everyday professional and educational tasks? 5. Results To verify the proposed hypotheses, a probit model was applied, which is suitable for analyzing a binary dependent variable - in this case, indicating whether the respondent reports a lack of cognitive independence when using artificial intelligence tools (1 = yes, 0 = no). In interpreting the results, the marginal effects (dy/dx) and statistical significance levels (p-values) are of key importance, as they indicate the strength and direction of the influence of the independent variables on the probability of cognitive dependence. Model functions: P(Y = 1 ∣ X )= Φ(β0+ β1 X1 + β2 X2 + ⋯ + β𝑘X𝑘 ) (1) Where: • Φ – the cumulative distribution function of the standard normal distribution • Y – the dependent variable (e.g., lack of independence = 1) 101 • Xk – the set of explanatory variables Table 1 presents the data on the dependent variables and the coding method. Table 1. Dependent variable (Y) Name Description Coding Lack of independence Does the user report being unable to complete tasks independently without AI 1 = yes; 0 = no Source: (author) Table 2 lists the independent variables and their coding scheme. Table 2. Independent variables (X) Name Description Gender (female) 1 = female, 0 = male Higher education 1 = bachelor's / engineer / master’s degree, 0 = vocational / secondary education City > 200,000 1 = yes, 0 = no Frequency of AI use scale 1–5 (the higher the score, the more frequent the use) Independence (scale) from 1 – full independence to 4 – lack of independence Abandonment of ideas scale 1–4 (1 = never, 4 = often) Creativity (enhancement) 1 = AI enhances creativity, 0 = does not Source: (author) The analysis of marginal effects presented in Table 3 indicates that the frequency of using AI tools has a significant and positive impact on the declared lack of independence. The marginal effect is +0.1217 (based on an earlier version of the model) with a p-value of 0.001, which means that the more frequently respondents use AI chatbots, the more likely they are to report being unable to complete tasks without such support. This result directly supports the hypothesis that intensive use of AI may contribute to the phenomenon of cognitive offloading, thereby limiting the ability to think independently. Table 3. Marginal effects of the probit model – dependent variable: “limited independence” Variable Marginal Effect (dy/dx) Standard Error (SE) z-value p-value 95% Confidence Interval (lower – upper) Gender (female) –0.0428 0.066 –0.652 0.514 –0.171 – 0.086 Higher education +0.0752 0.065 +1.160 0.246 –0.052 – 0.202 Large city (>200,000) +0.0197 0.066 +0.298 0.766 –0.110 – 0.149 Frequency of AI use +0.0740 0.038 +1.933 0.053 –0.001 – 0.149 Independence (scale) +0.0008 0.036 +0.023 0.982 –0.069 – 0.071 Source: (author) The frequency of AI chatbots use had a statistically significant impact on the studied phenomenon. The marginal effect was +0.0740 (p = 0.053), meaning that each one-point increase on the frequency 102 scale (e.g., from “once a day” to “several times a day”) increased the probability of lacking independence by 7.4 percentage points. Although this result is close to the 0.05 significance threshold, it should be treated as a signal of a potential negative effect of intensive AI usage. Other variables, such as gender, education level, place of residence, general assessment of independence, and the tendency to abandon one's own ideas, did not show a statistically significant impact on the reduction of independence in the final model presented in Table 3. However, it should be noted that the variable “abandonment of ideas” had a positive coefficient (+0.1215) in the earlier, initial version of the model, which may indicate a tendency to conform to AI suggestions. Creativity, as an effect of using AI - i.e., the belief that a chatbot “enhances creativity” – had a negative marginal effect of –0.2546 (p < 0.001) in the initial model estimation. This indicates that respondents who perceive artificial intelligence as a tool that supports creative thinking are less likely to lose their independence. It can therefore be assumed that positive, creative uses of AI serve as a protective factor against its potentially passive influence. 6. Discussion of results Research indicates that intensive use of AI-based chatbots such as ChatGPT may lead to a weakening of users’ cognitive independence and a decline in the need for active information processing. This phenomenon, known as cognitive offloading, involves shifting mental tasks to external sources, which in turn diminishes analytical and reflective abilities (Zhai, Wibowo, Li, 2024). Individuals who use chatbots frequently tend to show lower cognitive engagement and reduced inclination toward reflection (Dylan, 2025; Shao, Huang, 2024), often mechanically accepting system suggestions without further elaboration. Moreover, the communication style—especially when emotional or persuasive - increases users’ susceptibility to AI influence (Rese, Tränkner, 2023). However, a creative approach to AI tools acts as a protective factor. Users who treat AI as a source of inspiration rather than a substitute for thinking tend to maintain a higher level of intellectual autonomy (Vecchiarini, Somià, 2023; Noy, Zhang, 2023). Importantly, these effects occur regardless of age, gender, or education level - they are universal and cross-cultural in nature (Uppal, Hajian, 2025). The diagnosis is clear: uncritical use of AI may undermine entrepreneurial capacity, shifting responsibility for decisions from humans to algorithms. The ongoing development of Artificial General Intelligence (AGI) - capable of autonomous decisionmaking - poses new ethical challenges related to control and accountability. However, Lisník suggests that the concept of entropy and tools such as the Human Resources Ratio may support the design of AI systems that ethically enhance human development, rather than degrade human capabilities (Lysa, Lisník, 2019). For this reason, it is essential to promote digital competencies with a strong emphasis on critical and creative use of AI, as well as education in the ethics of emerging technologies. 7. Conclusions and action recommendations Based on the conducted research, the following conclusions have been formulated: • Frequent use of AI chatbots is associated with a greater tendency to avoid independent decision-making and cognitive responsibility. • Abandonment of one’s own ideas correlates with reduced initiative, users are more likely to rely on ready-made AI suggestions, which may weaken individual thinking strategies. 103 • A creative approach to AI acts as a safeguard against perceiving AI as a substitute rather than a source of inspiration, and it supports higher intellectual autonomy. • Dependence on AI appears to be universal - no significant influence of gender, education level, or place of residence was observed. In light of ongoing technological transformation, it is essential to develop adaptive mechanisms, primarily in the fields of education and ethics. It is recommended to strengthen practical entrepreneurship education, with an emphasis on soft skills, critical thinking, and conscious use of digital tools, aiming to foster a modern, independent, and resilient society ready to face the challenges of the AI era. According to the Author, a person becomes impoverished because they enrich others - allocating their time, energy, and potential not to the realization of their own ideas, but to supporting external projects that rarely bring true self-fulfillment. Meanwhile, the level of entrepreneurial development - understood not only as material wealth but as an attitude of active, conscious, and responsible shaping of reality - forms the foundation of who we are, what we possess, what we are capable of, the relationships we build, and the direction we take as individuals and as a society. Today, entrepreneurship is not a luxury, but a prerequisite for autonomy. Its development offers us a chance to reclaim agency. Technology, including AI, can support us, but if misused, it may weaken our will, deepen inequalities, foster dependency, lead to digital dementia, depression, or chronic procrastination, and give rise to new forms of social and psychological disorders. References Adel, A., Ashan, A., Davison C. (2024). ChatGPT Promises and Challenges in Education: Computational and Ethical Perspectives, Education Sciences, 14(8), 814, MDPI. https://doi.org/10.3390/educsci14080814 Barney, J. (2024). Does Chat GPT Improve Doctors’ Diagnoses? Study Puts It to the Test, https://news.med.virginia.edu/research/does-chat-gpt-improve-doctors-diagnoses-study-puts-it-to-the-test/ Bawa, M., Špirková, D. (2016). Importance of Internet of Things and Big Data in Building Smart City and What Would Be Its Challenges. Lecture Notes of the Institute for Computer Sci, Conference: Smart City 360°. DOI:10.1007/9783-319-33681-7_52 Bryant, A. (2023). AI Chatbots: Threat or Opportunity? Informatics, vol. 10, MDPI. https://doi.org/10.3390/informatics10020049 Dylan, E. (2025). AI tools may weaken critical thinking skills by encouraging cognitive offloading, study suggests. PsyPost. https://www.psypost.org/ai-tools-may-weaken-critical-thinking-skills-by-encouraging-cognitiveoffloading-study-suggests/ Global Entrepreneurship Monitor. (2025). GEM 2024/2025 Global report: Entrepreneurship reality check, GEM. Katrenčík, I., Zatrochová, M. (2018). 2017 – year of the cryptocurrency. In European Entrepreneurship Forum 2017. Eurozone: Evolution or Revolution? Proceedings of the 11th International Conference, Prague: Newton Academy, 64–72. Khusainov, R. V., Lisnik, A., Zatrochová, M., Babiuk, A. M., & Mashkov, K. Ye. (2023). Economic and legal bases of the Carpathian Euroregion development during the COVID-19 pandemic (Hungary, Slovakia, Poland, Ukraine). Journal of Innovation and Entrepreneurship, 12(45). Krawczyńska, J. (1930). Równouprawnienie i białe niewolnictwo. Bluszcz. Społeczno-literacki ilustrowany tygodnik kobiecy, (43), 3–4. Krusiec, K. (2018). Ekonomiczne czynniki wpływające na poziom życia – analiza porównawcza krajów Unii Europejskiej, [in:] Zeszyty Naukowe WSEI, (14), 176–199. Krusiec, K. M. (2019). Przedsiębiorczość jednostki w świetle rozwoju gospodarki 4.0 – świadomość zagrożeń, [in:] Zeszyty Naukowe WSEI, (15), 156-174. 104 Krusiec, K. M., Palej, A. (2024). Human capital as a component of the business model: threats from artificial intelligence and a pragmatic approach in the era of modern challenges, [in:] MANanažment a EKOnomika. Journal of MANagement and ECOnomics, MANEKO, 56-64. Kuperová, M., & Lisnik, A. (2023). Innovative forms of marketing and their impact on customer behaviour. In Economics, Management & Business 2023: Contemporary Issues, Insights and New Challenges, 536–546. Lisnik, A., & Kuperová, M. (2023). Ethical aspects of working with data. W M. Horák, R. Doucek, & M. Basl (Red.), IDIMT-2023: New Challenges for ICT and Management – 31st Interdisciplinary Information Management Talks, 357–364. Lisnik, A., Janíčková J., Zimermanová K. (2020). Biometric systems and their use in social networks. IDIMT-2020: digitalized economy, society, and information management. Linz. TRAUNER Verlag. Living conditions in Europe – Poverty and social exclusion. (2024). Eurostat. https://ec.europa.eu/eurostat/statisticsexplained/index.php?title=Living_conditions_in_Europe_-_poverty_and_social_exclusion Lysa, L., Jenčo, M., Droppa, M., Lisnik, A. (2019). Entropy – a tool for measuring the quality of human potential. Quality – Access to Success, 20(170), 94–99. Noy, S., Zhang, W. (2023). Experimental evidence on the productivity effects of generative artificial intelligence. Science, 381(6654), 187–192. http://dx.doi.org/10.2139/ssrn.4375283 Palej, A., Krusiec, K. M. (2024). Nie należy myśleć i zapominać, należy myśleć i działać, wówczas działanie stanowi świadectwo myślenia. [in:] Nauki społeczne wobec wyzwań współczesności – aspekty psychologiczne, politologiczne i pedagogiczne, StANS, 167–186. Rese, A., Tränkner, P. (2023). Perceived conversational ability of task-based chatbots, International Journal of Information Management 74(8):102699. DOI: 10.1016/j.ijinfomgt.2023.102699 Shao, Y., Huang, C. (2024). Using augmentation-based AI tool at work: A Daily Investigation of Learning-Based Benefit and Challenge. Journal of Management. https://doi.org/10.1177/01492063241266503 Uppal K., Hajian S. (2025). Students’ perceptions of ChatGPT in higher education: A study of academic enhancement, procrastination, and ethical concerns. European Journal of Educational Research, 14(1), 199–211. https://doi.org/10.12973/eu-jer.14.1.199 Vecchiarini, M., Somià T. (2023). Redefining entrepreneurship education in the age of artificial intelligence: An explorative analysis, The International Journal of Management Education, https://doi.org/10.1016/ j.ijme.2023.100879 World Economic Forum. (2025). Future of Jobs Report 2025: Insight Report, WEF, 18–19. Zhai, C., Wibowo, S., & Li, L. D. (2024). The effects of over-reliance on AI dialogue systems on students’ cognitive abilities: A systematic review. Smart Learning Environments 11, Article 28. 111 Fear of being exposed (β = 0.6422; p = 0.005; ΔP = +0.18) was the second significant predictor of honesty, particularly among older, working students. The effect varied culturally: highest in Ireland (strict EU norms), moderate in Poland (GDPR), and lowest in Ukraine (weak legal safeguards). Traits like availability, diligence, and leadership were common in CVs but poorly correlated with actual behaviour - suggesting that even moral attributes are being "aestheticized" via AI tools. The findings raise ethical concerns about authenticity and highlight the need for legal regulation, practical skill verification, and ethical recruitment standards across all countries studied. 7. Conclusions and recommendations The conducted study clearly confirmed the existence of discrepancies between the competencies declared in CVs and the actual skills of candidates. Among the many variables analyzed, only punctuality (ΔP = +0.24) and fear of being exposed (ΔP = +0.18) proved to be significant predictors of authenticity and real professional preparedness. This phenomenon aligns with the findings of Billieux et al. (2024), who highlight the growing role of AI tools in creating an "aestheticized" professional identity detached from genuine competencies. While artificial intelligence can facilitate the creation of application documents, it also becomes a tool for controlled dishonesty. Candidates often shift moral responsibility for CV content to algorithms, which - as noted by Müller and Kvalnes (2013) - leads to the erosion of trust between applicants and employers. At the international level, the study revealed a lack of clear regulations regarding the use of AI in the application process: • Poland: No specific provisions; only general references in GDPR, the AI Act, and the Labour Code. • Ireland: EU regulations (GDPR, AI Act) apply, with an emphasis on algorithmic transparency. • Ukraine: No comprehensive legal framework; high risk of systemic discrimination and digital exclusion. Recommendations: • Empirical verification of skills – implement practical workshops, assessment centers, and recruitment simulations. • Application education – integrate ethics of authenticity into CV creation processes. • Candidate potential assessment – apply entropy-based evaluation methods (Lysa & Lisnik, 2019). • Labeling AI-generated CVs – e.g., through a disclaimer such as: “Generated with the use of AI tools.” • Implementation of ethical standards – transparency, equal access, and fairness should be the foundation of new recruitment practices. • Reducing technological inequalities – combat digital exclusion, especially in countries with limited access to AI. In the face of growing “declaration inflation”, it is essential to shift the focus from form to substance - ensuring credible, practical confirmation of qualifications while upholding ethical and regulatory principles. 112 References Billieux, J., von Hammerstein C., Ciudad-Fernández, V. (2025). People are not becoming “AIholic”: Questioning the “ChatGPT addiction” construct. Addictive Behaviors, (166),108325. Burton, S. L. (2019). Grasping the cyber-world: Artificial intelligence and human capital meet to inform leadership. International Journal of Economics, Commerce and Management, 7, 707–759. Carlà, M. M., Gambini, G., Baldascino, A., et al. (2024). Large language models as assistance for glaucoma surgical cases: A ChatGPT vs. Google Gemini comparison. Graefe's Archive for Clinical and Experimental Ophthalmology, 262, 2945–2959. Gill, S. S., et al. (2024). Transformative effects of ChatGPT on modern education: Emerging era of AI chatbots. Internet of Things and Cyber-Physical Systems, (4), 19–23. IBBC Group. (2023). Kłamstwa w CV na polskim rynku pracy – raport z badania ankietowego. Rojewska, M. (2024). Jak przebiega rozmowa kwalifikacyjna w Polsce? Badanie. https://interviewme.pl/blog/pytaniana-rozmowie-kwalifikacyjnej-badanie Javaid, M., Haleem, A., Singh, R. P. (2023). A study on ChatGPT for Industry 4.0: Background, potentials, challenges, and eventualities. Journal of Economy and Technology, (1), 127–143. Katrenčík, I., & Zatrochová, M. (2018). 2017 – year of the cryptocurrency. In European Entrepreneurship Forum 2017. Eurozone: Evolution or Revolution? Proceedings of the 11th International Conference, 64–72. Prague: Newton Academy. Khusainov, R. V., Lisnik, A., Zatrochová, M., Babiuk, A. M., & Mashkov, K. Ye. (2023). Economic and legal bases of the Carpathian Euroregion development during the COVID-19 pandemic (Hungary, Slovakia, Poland, Ukraine). Journal of Innovation and Entrepreneurship, 12. Kuperová, M., Lisnik, A. (2023). Innovative forms of marketing and their impact on customer behaviour. In Economics, Management & Business 2023: Contemporary Issues, Insights and New Challenges, 536–546. Lisnik, A., Kuperová, M. (2023). Ethical aspects of working with data. IDIMT-2023: New Challenges for ICT and Management – 31st Interdisciplinary Information Management Talks, 357–364. Lisnik, A., Janíčková J., & Zimermanová K. (2020). Biometric systems and their use in social networks. IDIMT-2020: digitalized economy, society, and information management. Linz. TRAUNER Verlag. Lysa, L., Jenčo, M., Droppa, M., Lisnik, A. (2019). Entropy – a tool for measuring the quality of human potential. Quality – Access to Success, 20(170), 94–99. M. Bawa, D. Špirková, 2015. Importance of Internet of Things and Big Data in Building Smart City and What Would Be Its Challenges. Smart City 360o , Springer, 533-540. Müller, R., Andersen, E. S., Kvalnes, Ø. (2013). The interrelationship of governance, trust and ethics in temporary organizations. Project Management Journal, 44(4), 26–44. Putri, A. E., Dinata, W. W., & Basri, D. C. (2024). The influence of addiction to using artificial intelligence (AI) in Generation Z in guidance and counselling. BIC Proceedings, 159–164. Szczepanek A., ResumeLab. (2023). The Truth about Lying to Get a Job, https://resumelab.com/career-advice/lying-toget-a-job Yankouskaya, A., Liebherr, M., & Ali, R. (2024). Can ChatGPT be addictive? A call to examine the shift from support to dependence in AI conversational large language models. In Human-Centric Intelligent Systems. Springer, 2–19. 113 VIRAL MARKETING BEHAVIOUR OF GENERATION Z Martina Kuperová, Monika Zatrochová Slovak University of Technology in Bratislava Institute of Management [email protected], [email protected] DOI: 10.35011/IDIMT-2025-113 Keywords Digital marketing; Virtual marketing; Influencer marketing; Social media Abstract Customer preferences are changing with each generation as a consequence of technological advances. Today, Generation Z dominates, who are quite different from their parents. The aim of this paper is to highlight the innovative forms of marketing and their impact on customer behaviour, to describe in a selected segment of the target group the method of using viral marketing in a specific market segment. We obtained primary information using our own research. A questionnaire was used for the quantitative part of the research. We focused on the target group young people from 12-20 years old. In the context of post processing were used traditional scientific methods: method of analysis, synthesis, induction, deduction, comparison and observation. The output of the paper will be an evaluation of the impact of the use of social media and innovative forms on the buying behaviour of Generation Z. The aim of the contribution is to point out the usability of innovative forms of marketing and their impact on the purchasing behavior of young customers aged 12-20. Therefore, marketers have to adapt the form of their campaigns and focus on promoting products using innovative and creative marketing strategies to win new customers and also retain loyal ones. 1. Introduction Nowadays it is very easy for marketers to analyse what is the traffic on their website, what demographic group visits it the most. This information is very valuable and from this, further strategy can be developed to drive customers to purchase. The presentation of brands using modern social networks has greatly developed during the measures to protect society from the spread of Covid. This helped to develop social media marketing activities and thus changed the behavior of customers. Mobile applications and social media platforms have been claimed to be drivers of change in consumers’ behavior. In this vein, the confinement measures had an enormous impact on the way in which people purchased. The Internet and the online shopping made possible the continued purchase of many products and services, yet it changed substantially the customer journey map, urging companies to understand this new experience and to adapt accordingly (Vazquez-Martínez, 2021). 114 Monitoring the development of customer behavior on social networks is one of the important sources of information necessary for obtaining knowledge necessary for the development of marketing strategies of brands. Generation Z, defined as individuals born between 1995 and 2009, which is the youngest and largest consumer group for the period from 2017 to 2030. This generation are innovative, pragmatic, narcissism-oriented, and averse to negative events; they tend to have a strong focus on praise and a high level of social approval. Most importantly, they are known as technology-savvy digital natives, the first generation to be born into an entirely digital world, and thus they have grown up developing relationships with digital technologies extensively. From a marketing perspective, their great dependence on SNS plays a significant role in generating sales and revenue, especially under the economic disruption of the Covid-19 pandemic (Wang et al., 2023). Generation Z is known for its special and unique way of looking at values. One of the facts is lifestyle. It is influenced both by social networks, but also by the social environment in which the individual is located. It is very important to examine the relationship of the individual and his lifestyle in relation to values in the decision-making process in purchasing behavior. One of the modern ways of paying for their purchase are cryptocurrencies. They are also characterized by their digital nativity and openness to technological innovation, demonstrates a growing inclination toward using cryptocurrencies for purchasing goods and services, reflecting a shift in consumer behavior within emerging financial ecosystems (Katrenčík, 2020). Generation Z is constantly connected with others, initially it was a connection using phone calls, SMS or MMS messages, after the advent of social networks, the transfer of information has intensified even more. From the point of view of access and use of information technologies, we can consider Generation Z as one of the most informed generations. On the other hand, terms such as stress, burnout, frequent job changes, conflicts, lack of concentration appear more often in this generation than in others (Kajnaová, 2024). Research into brand positioning on the market shows that it is very necessary to focus on the right market segment and to communicate in a targeted manner, e.g. product packaging. There are many factors that influence the successful sale of a product. One of the basic factors is the name (name) of the product (Zajko, 2022). This factor is closely related to the study of the customer's purchasing behavior and the digital strategic marketing of the brand in close connection with advertising. 1.1 Viral marketing The goal of viral marketing is to reach the largest possible audience through social sharing. There are different ways of viral sharing, how through emails, videos, mobile apps. The main goal is to create a campaign or an advertisement that will be shared everywhere. This type of campaign is unique in that it does not use a paid distribution channel but relies on social sharing. If such content is shared by a celebrity or influencer, there is a high likelihood that it will go viral. As the potential of viral marketing began to become more and more apparent to companies, they struggled to understand the factors that influence a successful viral campaign. The STEPPS model describes six factors that are believed to contribute to the spread of content. Social Currency, if a user believes that sharing a piece of content will make them look favourable to their audience, they are more likely to share it. Triggers are anything that connects thoughts, ideas or different stimuli. For viral marketing to be effective, people should be able to identify with it, it should match the everyday experiences of consumers. Emotions, people react most to content that evokes an emotional response 115 in them. People who share such emotional content may feel that they are improving their status by showing interest in emotional, socially important issues. Public, if it is a message, advertisement or video that is shared by many individuals or various celebrities, the user is more likely to share something like this. Practical Value, an effective method of making content go viral is to provide some added value to those who share such content. People are happy to engage with something that promotes a good cause and are more likely to share such content. Stories, people are more likely to understand things better if they are presented to them in the form of a story. The story must be engaging, it must tend to motivate social media users to share it. Sharing an advertisement about a company, a product in a concise, visually appealing format can have a very good effect, which can make the brand more visible (Directive, 2023). Viral marketing is an area of marketing confronted with circulation, the need for commerciality, and most importantly, with the integrity of the content creator (Lisnik, 2024). The advantages viral marketing include: Rapid growth – a company's digital presence can grow through viral content. If a company has a larger reach, it gives it a good opportunity to attract new customers and also retain customers. Brand awareness – viral content helps a brand to introduce itself to customers and showcase its products. People will first remember a funny video which they will then associate with the brand. This kind of advertising is most beneficial for startups or small businesses. Cost effective – a good campaign will be effective and will not have a large cost as it relies on people sharing without the need to pay. Because of the ease of sharing content, there is no need to do extra work because content is shared naturally (Study Smarter, 2022). The disadvantages of viral marketing are: Negative advertising – the campaign can have a negative impact with users, which can cause negative advertising to spread, which can have a disastrous impact on the company. It may happen that customers stop buying the products of a given company because of such advertising, influencers stop working with them. Implementation and control – viral marketing is unpredictable, so there is a possibility that the ad may not reach the target group of users, making management and control more difficult. Wasted time – it may happen that the campaign does not generate interest among consumers. Feedback – gathering feedback can be challenging as it is viral advertising which cannot be measured in any way (Study Smarter, 2022). In the field of viral marketing, it is important to pay attention to the content of the information, including from the point of view of truthfulness. Since 1992, we have the first memory of the connotation of state law, which declares a minimum of due diligence for the truthfulness of the content of viral information. (Lisnik, Majerník, 2023). 1.1 Influencer marketing The most charismatic among us have always influenced what we buy. It used to be the Mary Kay Lady, the Avon Lady, today it's the popular people on social media Influencer marketing is a collaboration between an influencer and a brand, whereby it is the promotion of a product or service. The role of an influencer is to promote a product or service on their profile. Many times they have different discount codes or contests in collaboration with the brand. The brand chooses the influencer to collaborate with according to different criteria such as : target audience, life values, content on the profile, number of followers. Companies can collaborate with influencers to become brand ambassadors. An ambassador gets free products on a regular basis, also promotes them and has different discount codes on a regular basis (Geyser, 2023). This type of marketing is not without risks. It can be riskier and harder for big brands to appear authentic when working with an influencer, as it is always a monetary transaction between the 116 influencer and the company. When it's a lesser-known brand, the promotion may seem more authentic. It's also important to consider that influencer brand loyalty doesn't always have a positive effect. When a brand collaborates, it cannot influence social media behaviour and when influencers behave unprofessionally, it can jeopardise the brand's reputation. (McKinsey&Company, 2023) The scope and content of information shared on social networks also needs to be strictly protected, as it is largely information subject to the principles of the GDPR (Lisnik, Bretz, 2023). 1.3 Social media In 2006, one of the most popular social networking sites that has over a trillion users, Facebook, came into the world. Later in 2010 came Instagram where users could add photos, the platform came through many upgrades and now users can not only add photos or even chat there, try different filters but also follow their favorite celebrities or influencers. And last but not least, TikTok came to the social networking world in 2018, it is the most used social media of today, which has over three trillion users. Users post short videos of themselves dancing, painting, doing various tricks, etc. (Britannica, 2023). The great advantage of social media is the quick feedback from customers, they share their positive and negative experiences with products or services. To which companies can quickly respond and resolve negative customer reviews. Marketers are also using social media for crowdsourcing. Crowdsourcing is getting information, opinions from a large group of people who share information on social media (Edosomwan, 2011). Working with the content and meaning of words is a very important aspect of the information offered. Therefore, the content of the shared information also changes according to the conditions (circumstances) and the person interpreting the information. This phenomenon is especially highlighted on social networks (Ambrózy, 2023). Social media has different forms of platforms, these can be divided into four main categories: Social networks – users use social networks to communicate, share opinions or ideas. Users create a profile where they can share their age, gender, where they live, or they can choose to use such a platform to create different groups where people share their products, opinions, tips and tricks. Examples of such social networks include Facebook. Media sharing networks – users create different 'content' which they share. This content can be of different nature, such as: funny, educational, informative. Many companies use such media to share information about their products and their company. Such media include YouTube, Instagram, TikTok. Community social networks – this type is mainly used for discussions, users hold discussions on various topics. Review social networks – this type of social network is focused on customer reviews of a product or service. Individual reaction to the product is an important element in the creation of the social context of the products. Social networks thus create a model of relationship with the evaluator based largely on his feelings, experience, or validity (his social impact in the community). Individuals can take a series of actions on Facebook when engaging with a brand. They can like or react to a post, comment on another and share the other. Regarding reactions, an extension of the like button, users can express their sentiments (love, haha, wow, sad and angry) toward a publication. Furthermore, commenting provides users the opportunity to express their opinions through debate with others while sharing is considered the most valuable form of user engagement for brands, considering its potential for viral redistribution. Liking a post requires less commitment than commenting or sharing. Clicking is enough to like, while comments and shares demand additional actions and extra commitment (Romão, 2019). 117 1.4 TikTok TikTok, was first launched in 2016 in China. The app was very popular, so the parent company, ByteDance, introduced it to the international market. In 2017, with the acquisition of Musical.ly, the company was able to establish itself in the international market as it imported around 80 million users, most of which were from the US, onto the TikTok platform. In 2023, Tik Tok has just over 1.53 trillion active users, with just over 3 trillion people having downloaded the app, of which up to 57% are women. It is the world's sixth most popular medium. (IQBAL, Mansoor, 2023) Every day, a Tik Tok user spends 1.5 hours on the platform, for a yearly total of 547.5 hours. The most popular category is entertainment, followed by dancing, pranks, fitness, beauty, fashion, cooking. The engagement rate on Tik Tok in terms of micro-influencers is 17.96% as opposed to Instagram where it is only 3.86%. Larger influencers have an engagement rate of 4.96% on Tik Tok and 1.21% on Instagram. (Ruby, 2023) It's important to remember that with such a huge number of videos, you need to be creative, follow trends and also post videos on a regular basis. A clever way to break through on TikTok, for example, is a 'challenge', where you try to get as many users as possible to take part in the challenge, with users using a hashtag that lets other users know what the challenge is about. This is a very good opportunity for businesses to connect with their customers. Other popular videos are lifehacks, where users present interesting tutorials to followers. (Gašparová, 2022) 2. Methods The main objective of the present paper is to analyse the current state of the use of modern marketing methods, especially viral marketing in a specifically targeted group of respondents. The basic scientific methods such as observation, method of analysis, synthesis, induction, deduction, abstraction and comparison have been used in the preparation of the paper. The structure of the paper follows the structure of scientific papers: introduction, aim and methods, results and discussion, conclusion. At the first stage, we set the objectives of the paper, selected the relevant scientific methods to be used in the treatment of the chosen issue, then through the fulfilment of the partial objectives we get to the actual treatment of the issue and the assessment of the current situation. Based on the facts and facts dealing with the issue, we have identified the analysis of the results as an important factor, which we have formulated into conclusions, recommendations and evaluation of the objectives. The paper presents the results of our own primary research carried out by questionnaire method. The questionnaire was created using the Click4Survey website and was made available from February 25, 2023 to March 25, 2023. Respondents were approached through acquaintances, family friends, students, and lecturers. A total of 335 potential respondents opened the survey, but only 252 responded. 3. Results In cooperation with Herba Drug, s.r.o., we focused on social media and digital marketing research in the field of hair and body cosmetics oriented to Generation Z - using a questionnaire survey, which was oriented to the target group of 12-20 years old. They identified the following three sub-objectives: What social networks are most used by respondents. On what basis they choose a product. What external influences affect the purchase of products. 118 The questions in the questionnaire were also communicated with Herba Drug, Ltd (a cosmetics company dealing with hair and body cosmetics), these were modified and asked in such a way that the survey would be beneficial for them as well. In the questionnaire we asked what social networks the respondents use. A question where respondents could select multiple answers was what social networks they use the most. The most used social network is Instagram, which is used by 199 respondents. The second most used network is Tik Tok, which is visited by 95 respondents. The least used network is BeReal, which is a relatively new social network, so we consider this to be an expected result. Only 8 respondents use this network. In the next section, we asked what influences their purchase, if there is any viral video or advertisement that influenced them in hair and body cosmetics. The questionnaire was created using the Click4Surrvey website and was made available from February 25, 2023 to March 25, 2023. Respondents were approached through acquaintances, family friends, students and teachers. A total of 335 potential respondents opened the survey, but only 252 responded. Of the Slovak influencers, Moma is the most followed, followed by chemist, mom and cosmetics product development manager B_b_w_k and the same number of respondents answered Jana Hrmová and Radka Žilinčík. The most charismatic among us have always influenced what we buy. It used to be the Mary Kay Lady, the Avon Lady, today it's the popular people on social media. Of the foreign influencers, the most prominent are the Kardashians and Kylie Jenner, Selema Gomez, also footballers such as Cristiano Ronaldo and Robert Lewandowski. Another question in the survey asked what respondents choose cosmetics based on. Most respondents choose based on scent (65.1%), followed by price (50.8%). For 36.5% of respondents, recommendation from family and friends is important. The least important factor influencing the choice of cosmetics is advertising on TV and in the media, only 4.8% of respondents chose this answer. The next answer to the question if there is any viral video, advertisement, billboard in the field of cosmetics that influenced the respondents. 83.3% of the respondents answered no or did not answer at all. Head & Shoulders brand advertisement influenced 2.8% of the respondents, the reason was footballers who were doing advertisement for the brand. The Old Spice brand was also written by 2.8% of respondents, only one reason was given and that was an animal in the shower. Respondents also wrote other brands such as Adidas, Schauma, Nivea, Florence by Mills, Nature Box. 4. Discussion TikTok is currently the most used social network, where new users are added every day. Many companies abroad are using this medium to promote their brand. In Slovakia, this market is still new and unexplored. Few cosmetic companies use TikTok in Slovakia, there is a great potential here, because the competition is still very small. From our market research we found that only 3 brands out of 20 have a Slovak TikTok. All three are Slovak brands. Out of 252 respondents, TikTok is used by 95 respondents, which is 38%. The most used medium is Instagram, which is used by 199 respondents, which is 79%. Videos that are uploaded on TikTok are also very popular on Instagram now. The app introduced Instagram Reels in 2020, as an effort to rival the TikTok platform. The app has also added a separate section where users can watch these short videos. Many users share the same content on both social networks. Instagram Reels 119 is also a very good way for businesses to interact with their audience and promote their brand at the same time. In this way, they can showcase their products, share behind the scenes of production, and create various educational content, whether about their company, products, or ingredients. The length of a video on Instagram can be from 1590 seconds, while the length of a video on Tik Tok can be from 15 seconds to 10 minutes. We found that only three brands have a Slovak Tik Tok, and they are Biofy, Kvítok and Two cosmetics. Biofy currently has 905 followers, Kvítok has 44 followers, and Two cosmetics has 1080 followers. All three are natural brands. Most of the brands have TikTok, but only the foreign one. The fact that not many brands in Slovakia have this social network yet, so the space on Slovak TikTok with hair and body cosmetics remains open. And right now is a good opportunity to take a leading position on this social network. 5. Conclusion The paper is focused on identifying the state of use of current viral marketing methods in a specific group of respondents and in a specific market segment. Tik Tok is the most used viral tool for presenting videos especially among the young generation. Its main advantage is the "anatomical" positioning of the mobile device while viewing the video. It also changes the philosophy of artificially placing advertisements in videos which has caused an increase in the interest of respondents. Tik Tok is viewed by respondents on average twice as long as other viral assets. Young people of our respondents' age have abandoned traditional forms such as FB and are moving to viral networks such as Tik Tok and Instagram. The presentation of the survey results shows that Tik Tok has a high potential to improve the sales leads of a specific product line especially grounded to a specific group of respondents. A big advantage is the expansion of the offer and product innovation. It is also of great importance for collecting data on the basis of which we would be able to specify customer needs to satisfy them by developing and innovating products and product lines. The contribution of the presented article is the description of the current potential of using viral media and viral marketing for the presentation of specifically targeted products for generation Z. The paper also identifies channels for communicating information shared among a specific group of the population, which is absolutely different from previous generations, but forms a significant part of current and future buyers of products and services. identifying these channels and needs defines the future framework and means of marketing, especially viral marketing. Acknowledgement Scientific Paper was elaborated within the framework of the project KEGA č. 023STU-4/2023. References Ambrózy, M., Lisnik, A., Roubalová, M. (2023) Homiletic accents of ethical moments in the Protrepticus by Clement of Alexandria. In Acta Missiologica. Vol. 18, no. 1 (2024), s. 55-66. ISSN 1337-7515. Directive (2023) What is viral marketing? [online]. https://directiveconsulting.com/resources/glossary/viral-marketing/. Edosomwan, S., & Prakasan, S. K., & Kouame, D., & Watson, J., & Seymour, T. (2011). The history of social media and its impact on business. Journal of Applied Management and Entrepreneurship. [online, 2023-06-03]. https://www.proquest.com/scholarly-journals/history-social-media-impact-on-business/docview/889143980/se-2 120 Fernández, E., & Valle, S. (2019). Battle for dominant design: A decision-making model. In: European Research on Management and Business Economics. Vol. 25, Issue 2, p. 73. https://www.sciencedirect.com/science/article/pii/S2444883418302195, DOI: 10.1016/j.iedeen.2019.01.002 Gašparovská, M. (2022). TikTok as a great marketing opportunity in 2022. [online, 2024-09-04]. https://psdigital.sk/blog/tiktok-ako-marketingova-prilezitost-v-2022/ Geyser, W. (2023). What is Influencer Marketing? – The Ultimate Guide for 2023. [online, 2024-09-01]. https://influencermarketinghub.com/influencer-marketing/ IQBAL, Mansoor (2023). Tik Tok Revenue and Usage Statistics. [online, 2024-09-05]. https://www.businessofapps.com/data/tik-tok-statistics/ Kajanová, J., Papula, J. (2024) formuje generácia z inovačné procesy o vzdelávaní? = Is generation z generating innovative processes in education? [online, 2025-04-02]. In MANEKO Manažment a ekonomika podniku. (16)2. 146-156. ISSN 1338-5127. https://www.maneko.sk/pdf/articles/LAGP8796.pdf. DOI: 10.61544/mnk/LAGP8796 Katrenčík, I. (2020). Riadenie volatility na vysoko volatilných trhoch – kryptomeny = Volatility management in highly volatile markets – cryptocurrencies. [online, 2025-04-03]. MANEKO: Manažment a ekonomika podniku. 12(2), 172–180. ISSN 1337-9488. https://www.maneko.sk/casopis/pdf/2_2020.pdf Lisnik, A., Bretz, P., Warhol, M., Majerník, M. (2023) Specifics of Ethical Principles of System Data Protection in the Service Segment. In IDIMT-2023 : New challenges for ICT and management [elektronický zdroj]. 1. vyd. Praha: Prague University of Economics and Business, 2023, S. 373 - 380. ISBN 978-3-99151-176-2. DOI: 1035011/IDIMT.2023-373 Lisnik, A. Majernik, M. (2023) Aspects of Strategic Management and Online Marketing. In Developments in Information and Knowledge Management Systems for Business Applications [S.]. Schwitzerland : Springer Cham, 2023, S. 4980. ISBN 978-3-031-25694-3 Lisnik, A., Majerník, M., - Salanci, J. (2024) Ethical perspektives and the issue of authorial integrity in the use of AI in contemporary academic practice. In IDIMT-2024: Changes to ICT, Management, and Business Processes through AI: 32nd Interdisciplinary Information Management Talks. 1. vyd. Linz: TRAUNER Verlag, 2024, S. 405-409. ISBN 978-3-99151-527-2 McKinsey&Company (2023). What is influencer marketing. [online, 2023-06-01]. https://www.mckinsey.com/featuredinsights/mckinsey-explainers/what-is-influencer-marketing#/ Romão, M. T., & Moro, S., & Rita P., & Ramos P. (2019). Leveraging a luxury fashion brand through social media. In: European Research on Management and Business Economics. Vol. 25, Issue 1, p. 20. ttps://www.sciencedirect.com/science/article/pii/S2444883418301104. DOI: 10.1016/j.iedeen.2018.10.002 Ruby, D. (2023). 36 TikTok Statistics 2023: How many Users Are There! [online, 2023-06-06]. https://www.demandsage.com/tiktok-user-statistics/ Study Smarter. (2022). Viral marketing. StudySmarter UK. [online, 2023-06-03]. https://www.studysmarter.co.uk/ explanations/marketing/integrated-marketing-communications/viral-marketing/ Vázquez-Martínez, U. J., & Morales-Mediano, J., & Leal-Rodríguez, A. L. (2021, September-December). The impact of the COVID-19 crisis on consumer purchasing motivation and behavior. In: European Research on Management and Business Economics. Vol. 27, Issue 3. https://www.sciencedirect.com/science/article/pii/S2444883421000255. DOI: 10.1016/j.iedeen.2021.100166 Yanhong, W., & Yang, S., & Liu, D. (2023). The effect of social media influencer marketing on sustainable food purchase: Perspectives from multi-group SEM and ANN analysis. In: Journal of Cleaner Production, Vol. 416, p. 14-15. https://www.sciencedirect.com/science/article/pii/S0959652623020486. DOI: https://doi.org/10.1016/ j.jclepro.2023.137890 Wang, Z., & Yuan, R., & Luo, J., & Liu, M. J., & Yannopoulou N. (2023). Does personalized advertising have their best interests at heart? A quantitative study of narcissists’ SNS use among Generation Z consumers. In: Journal of Business Research. Vol. 165. https://www.sciencedirect.com/science/article/pii/S0019850122002838 ?via%3Dihub. DOI: 10.1016/j.jbusres.2023.114070 Zajko, M., Katrenčík, I., Majerník, Š. (2022). Riadenie malých a stredných podnikov (1st ed., 312 pp.). Bratislava: SPEKTRUM STU. ISBN: 978-80-227-5186-5 127 IDIMT 2023: Nové výzvy pre IKT a manažment - 31. interdisciplinárne prednášky o informačnom manažmente. s. 357 - 3632023 31. interdisciplinárne prednášky o informačnom manažmente: Nové výzvy pre IKT a manažment, IDIMT 2023Hradec Králové6. septembra 2023 až 8. septembra 2023Kód 191916 Kuperová, M. – Lisnik, A. (2023) Innovative Forms of Marketing and their Impact on Customer Behaviour. In Economics, Management & Business 2023: Contemporary Issues, Insights and New Challenges. Prešov: Presov Unviersity Press, 2023, S. 536-546. ISBN 978-80-555-3215-8. V databáze: WOS: 001250452400059. Lisnik, A. – Bretz, P. – Warhol, M. – Majerník, M. (2023) SPECIFICS OF ETHICAL PRINCIPLES OF SYSTÉM DATA PROTECTION IN THE SERVICE SEGMENT. In IDIMT-2023 New Challenges of ICT and Management. s. 377. ISBN 978-3-99151-176-2. In database: DOI: 10.35011/IDIMT-2023-373 Lisnik A., Kuperová M., ETHICAL ASPECTS OF WORKING WITH DATA, IDIMT Hradec Králové. 2023. [Online]: https://epub.jku.at/obvulioa/content/titleinfo/9026930/full.pdf Lisnik, A., Majerník, M., Salanci, J. (2024) ETHICAL PERSPECTIVES AND THE ISSUE OF AUTHORIAL INTEGRITY IN THEUSE OF AI IN CONTEMPORARY ACADEMIC PRACTICE. In IDIMT-2024 Changes to ICT, Management and Business Processes through AI. s. 408. ISBN 978-3-99151-527-2. In database: DOI: 10.35011/IDIMT-2024-405 Madzík a kolektív. Dobrá prax v oblasti spoločenskej zodpovednosti v prostredí vzdelávacej inštitúcie. - 2014. [Online]: https://www.researchgate.net/profile/Peter-Madzik/publication/320419725_Dobra_prax_v_oblasti_spolocenskej_ zodpovednosti_v_prostredi_vzdelavacej_institucie/links/59e4a8c5458515393d611190/Dobra-prax-v-oblastispolocenskej-zodpovednosti-v-prostredi-vzdelavacej-institucie.pdf Mill, John Stuart. (2003). Utilitarizem ali O svobodi. Ljubljana: Krtina, 2003, ISBN - 961-6174-58-4 Mucha, B. – Kovaříková, B. – Kuperová, M. (2024). Finančná gramotnosť ako kľúčová kompetencia pre rozvoj kritického myslenia študentov vysokej školy = Financial literacy as a key competence for the development of critical thinking of university students. In MANEKO Manažment a ekonomika podniku. Špeciálne číslo (2024), s. 67-73. ISSN 1337-9488, 1338-5127. In database: DOI: 10.61544/mnk/QDQR6984. Zatrochová, M. – Kuperová, M. (2020) Vision and direction of the chemical industry in the environment 4.0. In Business environment horizons. Part 1 (elektronický zdroj): proceedings of the international scientific conference. 1. Vyd. Praha: Wolters Kluwer, 2020, s.217-223. ISBN 978-80-7676-043-1 128 DATA, AI AND DIGITAL-DRIVEN TRANSFORMATION: SHAPING SUSTAINABLE DIGITAL FUTURES 131 OPEN DATA READINESS: EXPLORATORY STUDY OF SLOVENIAN ENTERPRISES Mirjana Kljajić Borštnar, Andreja Pucihar University of Maribor Faculty of Organizational Sciences [email protected], [email protected] DOI: 10.35011/IDIMT-2025-131 Keywords Open data; readiness; maturity; enterprises, Slovenia Abstract Open data (OD) has emerged as a key driver of transparency, innovation, and economic growth. While governments have invested heavily in OD infrastructures, the actual adoption and integration of OD by private enterprises remain limited and insufficiently studied. This paper explores the OD maturity of Slovenian enterprises using a custom multi-criteria decision model based on the DEX methodology. Drawing from established theoretical frameworks and a systematic review of the literature, the study identifies organizational, technological, and environmental factors influencing OD use. A self-assessment of 28 Slovenian enterprises reveals that although awareness of OD is increasing, implementation varies significantly. These findings highlight key barriers, including lack of infrastructure, limited strategic alignment, and insufficient awareness among assessed enterprises. The DEX-based model proves effective in assessing OD readiness and offering tailored recommendations. This study contributes to understanding enterprise-level OD adoption and offers practical and policy-relevant insights to support further development of data-driven innovation in the private sector. 1. Introduction Over the past two decades, open data (OD) initiatives have gained significant momentum worldwide, aiming to make data collected by public sector institutions available for reuse. In 2003, the European Commission adopted the first Directive on the re-use of public sector information (PSI) (European Commission, 2003), which Slovenia transposed into national legislation the same year (Republika Slovenija, 2003). These legislative developments laid the foundation for the systematic sharing of data by public institutions, with the broader objective of encouraging private sector entities to also open and utilize data. Today, OD—encompassing both open government data (OGD) and data voluntarily shared by private organizations—is made accessible through public portals and governed by frameworks such as the 8 Principles of Open Government Data (Open Government Working Group, 2007) and Berners-Lee’s Five-Star Linked Data model (2012). According to the European Commission’s report based on the European Data Market Monitoring Tool (European Commission, 2020), the data market in EU27 and the UK reached a value of €75 billion in 2019, reflecting a 4.9% growth compared to the previous year. This highlights the growing 132 economic significance of data and the need for enterprises to be capable of capitalizing on such market developments. EU institutions and national governments continue to allocate substantial resources toward the development of open data infrastructures and services. A key priority is to obtain feedback on how these resources are being utilized and to promote greater adoption of OD by private organizations. However, despite these investments, the actual use of OD by enterprises remains insufficiently explored. While OD is expected to contribute to transparency, civic engagement, and economic innovation, it is still unclear to what extent businesses are aware of, utilize, and are organizationally prepared to integrate OD into their operations, decision-making, and innovation processes. The primary objective of this study is to explore the open data maturity of Slovenian enterprises. To evaluate OD maturity, we developed a multi-criteria decision model based on the DEX methodology, which has been successfully applied in various real-world decision-making contexts, including digital readiness assessments of SMEs (Kljajić Borštnar & Pucihar, 2021). The model enables enterprises to conduct a structured self-assessment of their OD maturity, and based on the results, receive tailored recommendations to support further development. This paper presents preliminary findings based on the self-assessments of 28 Slovenian enterprises using the DEX model. Specifically, we report on their overall OD maturity levels, organizational and technological capabilities, the extent of data integration, the strategic use of data, and the types of data applied in decision-making. These early insights contribute to a better understanding of OD use among Slovenian enterprises and are relevant for both academic and policy providers, particularly in the context of ongoing governmental efforts to promote OD provision and use. 2. Literature Review Research on open data (OD) adoption has predominantly focused on user perspectives, employing established models such as the Unified Theory of Acceptance and Use of Technology (Saxena & Janssen, 2017; Shao, 2023; Talukder et al., 2019), the Technology Acceptance Model (Weerakkody et al., 2017), and Social Cognitive Theory (Wang, 2020). These studies consistently identify ease of use, perceived usefulness, and social influence as key drivers of OD adoption. Additionally, motivations for OD use have been explored across various sectors, including public institutions, private organizations, and entrepreneurs (Alawadhi et al., 2021; Mustapa et al., 2022; Wang & Lo, 2020). Several studies examined how user needs shape OD publication practices and the role of platforms in addressing public problems (Ruijer et al., 2020; Shepherd et al., 2019). OD has been shown to generate benefits such as innovation, efficiency, and transparency (Apanasevic, 2021; Jetzek et al., 2014; Shao, 2023). Research has also highlighted how OD supports democratic processes (Ruijer & Martinius, 2017), impacts organizational innovation (Huber et al., 2020), and enhances public administration (McBride et al., 2019; Wilson & Cong, 2021). Some studies have proposed measuring OD's impact through the added value it generates throughout its lifecycle (Attard et al., 2016; Magalhaes & Roseira, 2020). OD policy has been another key area of investigation. Researchers have proposed maturity levels, classifications, and analytical frameworks for OD policy evaluation (Attard et al., 2015; Ruijer & Meijer, 2020). Comparative studies identified variation in policy effectiveness across countries (Zuiderwijk & Janssen, 2014; Zuiderwijk et al., 2021), while recent work emphasizes designing policies that support OD reuse and institutional integration (Zhou et al., 2023). Initially, OD adoption was largely driven by government mandates, but over time, the responsibility shifted toward publishers and users. The European Union’s OD innovation use cases report (Granell 133 et al., 2022) aimed to accelerate adoption. Despite this, underutilization of OD remains a challenge. Researchers have documented technical, organizational, and legal barriers (Çaldağ & Gökalp, 2023; Sugg, 2022), and proposed strategies to overcome them (Huber et al., 2020; Wang et al., 2019). The quality of OD is another critical factor influencing its usability. Studies emphasize the importance of maintaining high-quality data, including its governance and continuous improvement (Ham et al., 2019; Schultz & Kempton, 2022; Moradi et al., 2022). As OD research has expanded, so has the need for categorization. Several frameworks and classification approaches have been proposed (Safarov et al., 2017; Cruz & Lee, 2016; Ansari et al., 2022), revealing two dominant research streams: one focused on government initiatives, and the other on regional adoption, barriers, and use cases (Ferencek et al., 2022; Wirtz et al., 2022). Finally, research into OD maturity models has gained traction. While several models exist for assessing publication readiness (Dodds & Newman, 2015; Mustapa et al., 2022), few address the maturity of organizations in terms of OD usage and integration into their operations (Çaldağ & Gökalp, 2022). This highlights a gap in evaluating OD as a strategic asset within enterprise processes. To address this gap, we have systematically examined the existing literature and identified factors that inhibit or encourage the use of open data by enterprises. We used Technology-OrganizationEnvironment framework to group the factors (Awa, Ojiabo, & Orokor, 2017), which was basis for further multiple-attribute model development. 2. Methodology This research is based in the Design Science Research (DSR) methodology (Hevner et al., 2004) where a solution to a real business problem is addressed by designing an IT artifact following a rigorous scientific process. The methodology comprises three main phases (Hevner et al., 2004): the rigor phase, in which the foundations for the research are established, such as existing theories, frameworks, constructs, etc., and determines the methodologies to achieve the most viable results. The relevance phase refers to business needs, which are addressed from human, organizational, and technological perspectives; and the design phase, during which the development and evaluation of an artifact are conducted. Within the rigor phase we conducted a systematic literature review to identify the theories employed to explain the motivation driving individuals and enterprises to use new technologies such as OD. We also reviewed the methods used to measure various aspects of the concept, as well as the resulting constructs, frameworks, and models proposed. Most importantly, we have identified factors that influence the use of OD in enterprises, which were used to design the multiple criteria decision model. We used the qualitative hierarchical multiple criteria decision methodology DEX (Bohanec & Rajkovič, 1990). The model is implemented as self-assessment on-line tool, that provides enterprises with insights into their strengths and weaknesses in their organizational preparedness for using OD and offer guidance to enhance their maturity level. The self-assessment tool offers enterprises insight into the potential benefits that OD could present to their business. Additionally, after a significant number of self-assessments conducted by enterprises, governments would gain insight into the state of the population of enterprises’ familiarity with the OD concept. 134 2.1 The model The model serves as a decision-making aid for alternative assessment and analysis. DEX method is utilized in a publicly available DEXi software which offers several analyses: interactive visual analysis, “what-if” analysis, pluses and minuses analysis, and others. Initially we have identified 118 driving and inhibiting factors, which we further refined through a facilitated brainstorming session with 16 domain experts using innovative brainstorming software and generative AI. In this process the group of experts generated 260 ideas, which were categorized into 11 groups and prioritized through voting. A smaller decision-making group combined the factors and ideas, which resulted in 56 attributes used in the final model. The final model consists of three subtrees representing the two basic dimensions: 1) organizational maturity and 2) business value. The two sub-trees are further subdivided into groups of attributes corresponding to the dimensions describing organizational maturity (organizational capability, technological capability and business environment) and business value (strategic value and competitive advantage). Each group is further divided. There are 32 basic and 24 aggregated attributes in the tree. For each attribute in the tree, the corresponding qualitative domain value is defined (for example: Autonomy {no, some, full} and the aggregation rules for inferring aggregated attribute values (for example: [if Autonomy {no} and Open communication {no} than Innovation culture {low}]). The model thus provides a composite value for each aggregated attribute in the tree, which is easily understood by the user and can be explained. The maturity of open data use is represented by a five-level qualitative scale: {lagging, basic, initial, advanced, winner}. This scale offers sufficient granularity to capture the nuanced differences in how enterprises utilize open data. 3. Results We present preliminary results of the assessment of maturity of enterprises to use open data conducted from September 2024 to March 2025. In total 28 enterprises from Slovenia had completed the assessment of maturity to use open data using DEX model implemented as a self-assessment online tool, 7 of those were from the East and 21 from the West cohesion region. The sample consists of 14 micro, 5 small, 3 medium and 6 large enterprises (Table 1). Most of the enterprises came from the Information and communication sector (6), Other service activities (5), Professional activities (4), Education (3), Manufacturing (3), Financial and insurance activities (2), and one from the Accommodation and food service activities, Agriculture, forestry and fishing, Wholesale and retail trade, repair of motor vehicles each. The overall maturity to use open data shows that only 2 enterprises reached the highest level of maturity (winner), 10 enterprises reached the level advanced, 8 reached the initial level, 7 the basic level and only 1 the lowest level (lagging). The overall maturity consists of three sub-categories: organizational capability, technological capability and business environment. Organizational capability evaluates the organization's capabilities in the areas of human resource management, organizational culture, governance, and strategic planning. 12 of the assessed enterprises reached the medium level of organizational capability and 11 of them excellent level. Only two enterprises reached low level of organizational capabilities. 135 The extent of technological capabilities refers to the maturity of data technologies, and the strategic role of information technology within the organization. Majority of the assessed enterprises falls into medium level (11), while there are 4 enterprises assessed as good and 8 as excellent. Only four of the enterprises fall into low level of technological capability. Only 4 out of 28 enterprises reported that they do not have their data organized, 5 of them uses data to support their daily activities, and 13 of them reported that their data is systematically structured to support the execution of daily business activities and to serve as a basis for informed decision-making. Finally, 6 of them reported that they regard data as a strategic asset for increasing added value and is therefore managed comprehensively throughout its entire lifecycle. Only 4 enterprises reported to rely on instinctive decision-making without systematically relaying on any type of data. Four reported to base their decisions on internal data, while 10 use internal as well as external data sources, and finally 10 out of 28 uses internal, external and open data sources in their decision-making processes. 4. Discussion and conclusion The findings of this exploratory study provide valuable initial insights into the open data (OD) maturity of Slovenian enterprises. The study confirms that while awareness and interest in OD exist among Slovenian organizations, actual implementation and integration of OD into strategic and operational processes remain uneven and, in many cases, limited. Only two out of 28 assessed enterprises achieved the highest maturity level ("winner"), while the majority were classified as "advanced" or "initial." This distribution indicates a growing but still emerging readiness for OD adoption. A deeper look at organizational capabilities revealed encouraging results, with a significant number of enterprises exhibiting strong internal foundations, such as human resource management and strategic planning. However, technological capabilities were more varied, with some organizations lacking the necessary infrastructure or strategic IT alignment to fully leverage OD. Moreover, while a third of enterprises view data as a strategic resource, a considerable number still rely on ad hoc or internal data only, with minimal integration of external or open data sources. This highlights a key maturity gap: although some organizations recognize the potential of OD, many have yet to institutionalize its use in decision-making and innovation. These findings align with existing literature that identifies barriers to OD adoption, such as limited technical resources, lack of awareness, unclear legal frameworks, and cultural aversion (Çaldağ & Gökalp, 2023; Sugg, 2022). They also reinforce the notion that enterprises, especially SMEs, require targeted support—including tools, training, and policy incentives—to increase their OD maturity. The DEX-based self-assessment model developed in this study has proven effective in identifying organizational strengths and weaknesses in relation to OD adoption. Its ability to provide tailored recommendations adds practical value and enhances its utility for both enterprises and policymakers. This study is subject to several limitations. First, the sample size is relatively small (n = 28), and while it includes a diverse mix of enterprises by size and sector, it may not fully represent the broader Slovenian enterprise population. Second, the data is self-reported, which may introduce subjectivity or social desirability bias in the assessment responses. Third, the model focuses primarily on internal organizational factors and may not fully capture external influences such as legal or ecosystem-level constraints. 136 Future research should focus on validating the model with a larger and more diverse sample, ideally across multiple countries or regions, to enable comparative analysis. Longitudinal studies could provide insight into how OD maturity evolves over time and in response to policy interventions or technological change. Further development of the model could also integrate external environmental factors and examine the interaction between OD maturity and enterprise performance outcomes, such as innovation rates or market growth. Additionally, future studies could explore sector-specific OD readiness, particularly in industries where data-driven innovation holds high potential (e.g., healthcare, energy, finance). Another avenue would be to examine the role of intermediary organizations or platforms in facilitating OD adoption among enterprises, especially SMEs. This study contributes to the growing body of research on open data adoption by addressing the underexplored area of enterprise readiness. Through a novel multi-criteria decision support model, it provides a practical framework for assessing OD maturity and offers actionable insights for organizations aiming to enhance their use of open data. The preliminary findings underscore the importance of strengthening both technological infrastructure and organizational awareness to support broader and deeper integration of OD in the private sector. Ultimately, this research lays the foundation for further empirical investigation and policy development to foster a more data-driven economy. Acknowledgement This research was funded by Slovenian Research and Innovation Agency and Ministry of Digital Transformation of Republic of Slovenia, grant number V5-2356, and Slovenian Research and Innovation Agency, grant number P5-0018. References Alawadhi, N., Al Shaikhli, I., Alkandari, A., & Kalaie Chab, S. (2021). Business Owners’ Feedback toward Adoption of Open Data: A Case Study in Kuwait. Journal of Electrical and Computer Engineering, 2021, 1–9. https://doi.org/10.1155/2021/6692410 Ansari, B., Barati, M., & Martin, E. G. (2022). Enhancing the usability and usefulness of open government data: A comprehensive review of the state of open government data visualization research. Government Information Quarterly, 39(1), 101657. https://doi.org/10.1016/j.giq.2021.101657 Apanasevic, T. (2021). Socio-economic effects and the value of open data: A case from Sweden. 23rd ITS Biennial Conference, 2021. http://hdl.handle.net/10419/238004 Attard, J., Orlandi, F., & Auer, S. (2016). Value Creation on Open Government Data. 2016 49th Hawaii International Conference on System Sciences (HICSS), 2605–2614. https://doi.org/10.1109/HICSS.2016.326 Attard, J., Orlandi, F., Scerri, S., & Auer, S. (2015). A systematic review of open government data initiatives. Government Information Quarterly, 32(4), 399–418. https://doi.org/10.1016/j.giq.2015.07.006 Awa, H. O., Ojiabo, O. U., & Orokor, L. E. (2017). Integrated technology-organization-environment (T-O-E) taxonomies for technology adoption. Journal of Enterprise Information Management, 30(6), 893–921. https://doi.org/10.1108/JEIM-03-2016-0079 Berners-Lee, T. (2012, January 22). Linked data. https://5stardata.info/en/ Bohanec, M., & Rajkovič, V. (1990). DEX: An Expert System Shell for Decision Support•. Sistemica, 1990(1), 145–157. Çaldağ, M. T., & Gökalp, E. (2022). The maturity of open government data maturity: A multivocal literature review. Aslib Journal of Information Management, 74(6), 1007–1030. https://doi.org/10.1108/AJIM-11-2021-0354 Çaldağ, M. T., & Gökalp, E. (2023). Understanding barriers affecting the adoption and usage of open access data in the context of organizations. Data and Information Management, 100049. https://doi.org/10.1016/j.dim.2023.100049 143 in 115 businesses. We then calculated the difference between actual utilization and future managers` expectations as ∆FA. The total sum of relative shares (AI_IT_rel) speaks to AI-driven manufacturing technology's actual use and future use. Table 2. AI-driven manufacturing technologies utilization AI_IT_abs AI_IT_absA AI_IT_relA AI_IT_absF AI_IT_relF ∆FA AI_IT_rel RPA 67 24 20.87 43 37.39 16.52 58.26 DT 82 31 26.96 51 44.35 17.39 71.30 PMS 63 51 44.35 12 10.43 -33.91 54.78 CVQC 41 21 18.26 20 17.39 -0.87 35.65 CR 62 23 20.00 39 33.91 13.91 53.91 SCOS 48 37 32.17 11 9.57 -22.61 41.74 AGV 61 33 28.70 28 24.35 -4.35 53.04 NLP 31 9 7.83 22 19.13 11.30 26.96 GD 67 46 40.00 21 18.26 -21.74 58.26 3DP 87 54 46.96 33 28.70 -18.26 75.65 Source: (authors) As Table 2 shows, the most used AI-driven manufacturing technologies are currently 3DP (46.96%), PMS (44.35%), GD (40.00%), and SCOS (32.17%). It answers the RQa: Which AI-driven manufacturing technologies are most widely used today? Figure 1. Prediction of the future AI-driven technologies utilization in total [%] Source: (authors) The most expected AI-driven manufacturing technologies in the future are DT (17.39%), RPA (16.52%), CR (13.91%), and NLP (11.30%). Expectations are calculated as the difference between the current and future use of AI-driven manufacturing technology. This answers the research question RQb: Which AI-driven manufacturing technologies are most anticipated by managers in the future? The technologies with the most negligible growth potential, as they are already sufficiently used today, are PMS (33.91%), SCOS (22.61%), GD (21.74%), and 3DP (18.26%). This answers the research question RQc: Which AI-driven manufacturing technologies have the smallest potential for deployment in a sample of large manufacturing companies? However, if we add up the relative shares of current use and future use, Figure 1 shows which AI-driven technologies will be most used in the future. The research revealed clear patterns in the adoption and expected use of AI-driven manufacturing technologies. Among the ten examined, 3D Printing, Predictive Maintenance Systems, Generative Design, and Supply Chain Optimization Systems were identified as the most commonly used in the surveyed companies. These technologies are likely to lead due to their established practical benefits, relatively mature development, and tangible impact on operational efficiency. For instance, 58,26 71,30 54,78 35,65 53,91 41,74 53,04 26,96 58,26 75,65 0,00 10,00 20,00 30,00 40,00 50,00 60,00 70,00 80,00 RPA DT PMS CVQC CR SCOS AGV NLP GD 3DP 144 3D Printing supports prototyping and customized production, Predictive Maintenance reduces downtime and maintenance costs, Generative Design enhances product innovation, and Supply Chain Optimization helps manage complexity and improve logistics performance. In contrast, technologies like Digital Twins, RPA, and NLP (though not yet widespread) show the highest increase in expected adoption. This trend reflects an interest in more advanced, integrated, intelligent systems. Digital Twins, for example, allow virtual replication of fundamental processes, enabling predictive analytics and real-time control. Their relatively low current use, paired with high future expectations, suggests growing awareness and improving readiness for implementation. Similarly, RPA is gaining attention as companies seek to automate repetitive, rule-based tasks, often in administrative or quality control processes. On the other hand, technologies such as Predictive Maintenance and 3D Printing show negative growth potential. This indicates that many companies that intended to implement them may have already done so, reaching a saturation point. These technologies have become standard tools rather than emerging innovations, especially in industries with high operational demands and mature digital infrastructures. These findings offer several practical implications for manufacturing enterprises. Companies can prioritize investment in Digital Twins, RPA, and Collaborative Robots to align with industry trends and remain competitive. Firms that have not yet adopted Predictive Maintenance or 3D Printing may need to assess why peers have integrated them successfully. Managers should also consider training and workforce adaptation strategies to support the transition toward these advanced technologies. The research also highlights several limitations and barriers. First, the study itself is based on a partial dataset. Although the sample size is statistically representative, additional data could provide more robust insights and enable deeper analysis across specific sectors. Moreover, the focus on large companies may not fully reflect the experiences and challenges faced by small and medium-sized enterprises, which often lack the same resources and technological infrastructure. From the adoption perspective, several practical barriers persist. Implementation complexity, integration with legacy systems, high initial investment costs, and a shortage of skilled personnel continue to slow progress, especially for advanced solutions like Digital Twins and NLP. Concerns around cybersecurity and data governance also remain unresolved in many companies. Furthermore, the organizational mindset is a non-trivial factor: even when technologies are available and technically viable, resistance to change and limited digital maturity can delay or derail implementation. 5. Conclusion This study offers a focused analysis of the current and expected use of AI-driven technologies in Slovak manufacturing, based on a survey of managers from large industrial companies. It explores ten specific technologies: Predictive Maintenance Systems, 3D Printing, Generative Design, Supply Chain Optimization Systems, Digital Twins, Robotic Process Automation (RPA), Collaborative Robots (Cobots), Computer Vision for Quality Control, Automated Guided Vehicles (AGVs), and Natural Language Processing (NLP). These technologies collectively form the backbone of AI-driven manufacturing transformation under the Industry 4.0 framework. The results suggest a maturing landscape where some technologies—particularly Predictive Maintenance, 3D Printing, Generative Design, and Supply Chain Optimization—have already seen broad adoption. These systems are now considered part of the operational core in many companies, contributing to greater efficiency, reduced downtime, and enhanced customization. However, their saturation also implies limited room for further expansion. Conversely, technologies such as Digital Twins, AI-enhanced RPA, Cobots, and NLP show strong potential for future deployment. These tools offer deeper integration, real-time responsiveness, and improved human-machine interaction, aligning with the evolving demands of intelligent, adaptive production environments. The findings suggest that future success will hinge not 145 only on adopting new technologies but also on overcoming these organizational, technical, and human barriers. Companies must invest in workforce development, change management, and robust digital infrastructure to fully realize the benefits of AI-driven manufacturing. The technologies with the greatest growth potential will likely be those that offer not just innovation, but ease of integration, scalability, and clear returns on investment. In conclusion, while the road to full AI integration in manufacturing is still under construction, this study provides a directional map. It underscores that true transformation is a multidimensional effort, requiring not only technology adoption but strategic alignment, cultural readiness, and sustained investment. As new research expands the dataset and more companies move from pilot phases to scaled deployment, a clearer picture will emerge of how AI will ultimately redefine manufacturing performance. Acknowledgment The paper is supported by the KEGA project no. 023STU-4/2023 Teaching financial literacy using information technology as a means of increasing the quality of life and eliminating the effects of the economic crisis. The authors hereby declare that they used the generative artificial intelligence ChatGPT 4.0 only to verify and correct the syntax of their written text, as they are not native speakers. The generative artificial intelligence was not used for research purposes, data analysis, interpretation, conclusions, discussion, or results. References Antosz, K., Pasko, L., & Gola, A. (2020). The Use of Artificial Intelligence Methods to Assess the Effectiveness of Lean Maintenance Concept Implementation in Manufacturing Enterprises. Applied Sciences, 10(21), 7922. https://doi.org/10.3390/app10217922 Bhatti, G., Mohan, H., & Singh, R. R. (2021). Towards the future of smart electric vehicles: Digital twin technology. Renewable and Sustainable Energy Reviews, 141(110801). https://doi.org/10.1016/j.rser.2021.110801. Eulerich, M., Waddoups, N., & Wagener, M. (2023). The dark side of robotic process automation (RPA): Understanding risks and challenges. Accounting Horizons, 38(2), 1-10. https://doi.org/10.2308/HORIZONS-2022-019 Fernandez, D., & Aman, A. (2021). Planning for a successful Robotic Process Automation (RPA) project. Journal of Information and Knowledge Management, 11(1), 103-117. Fuller, A., Fan, Z., Day, C., & Barlow, C. (2020). Digital twin: Enabling technologies, challenges, and open research. IEEE Access, 8, 108952-108971. https://doi.org/10.1109/ACCESS.2020.2998358. Hassan, N. M., Hamdan, A., & Shahin, F. (2023). An artificial intelligent manufacturing process for high-quality lowcost production. International Journal of Quality & Reliability Management, 40(7), 1777-1794. https://doi.org/10.1108/IJQRM-07-2022-0204 Hosamo, H. H., & Imran, A. (2022). A Review of the Digital Twin Technology in the AEC‐FM Industry. Advances in Civil Engineering, 2022(2185170), 1-17. https://doi.org/10.1155/2022/2185170 Hofmann, P., Samp, C., & Urbach, N. (2020). Robotic process automation. Electronic Markets. 30, 99–106. https://doi.org/10.1007/s12525-019-00365-8 Hyun, Y. G., & Lee, J. Y. (2018). Trends analysis and future direction of business process automation, RPA in the times of convergence. Journal of Digital Convergence, 16(11), 313-327. https://doi.org/10.14400/JDC.2018.16.11.313 Javaid, M., Haleem, A., & Singh, R. P. (2022). Artificial intelligence applications for industry 4.0: A literature-based study. Journal of Industrial Information Integration, 7(1), 83-11. https://doi.org/10.1142/S2424862221300040 Kumar, S. P. L. (2017). State of the art-intense review on artificial intelligence systems application in process planning and manufacturing. Engineering Applications of Artificial Intelligence. 65, 294-329. https://doi.org/10.1016/j.engappai.2017.08.005. Liu, M., Fang, S., Dong, H., & Xu, C. (2021). Review of digital twin about concepts, technologies, and industrial applications. Journal of Manufacturing Systems, 58(B), 346-361. https://doi.org/10.1016/j.jmsy.2020.06.017. 146 Madakam, S., Holmukhe, R. M., & Jaiswal, D. K. (2019). The future digital work force: Robotic process automation (RPA). Journal of Information Systems and Technology Management, 16(1), e201916001. https://doi.org/10.4301/S1807-1775201916001 Madni, A. M., & Lucero, S. D. (2019). Leveraging digital twin technology in model-based systems engineering. Systems. 7(1), 7. https://doi.org/10.3390/systems7010007 Mamede, H. S., Costa, D. A. S., & Oliveira, P. (2022). Robotic Process Automation (RPA) adoption: a systematic literature review. Management in Production Engineering, 14(2), 1-12. https://doi.org/10.2478/emj-2022-0012 Mashaly, M. (2021). Connecting the twins: A review on digital twin technology & its networking requirements. Procedia Computer Science. 184, 299-305. https://doi.org/10.1016/j.procs.2021.03.039. Mihaita, A. S., & Peimbert-García, R. E. (2022). Digital twin technology challenges and applications: A comprehensive review. Remote Sensing, 14(6), 1335. https://doi.org/10.3390/rs14061335 Santos, F., Pereira, R., & Vasconcelos, J. B. (2020). Toward robotic process automation implementation: An end-to-end perspective. Business Process Management Journal, 26(2), 405-420. https://doi.org/10.1108/BPMJ-12-2018-0380 Qi, Q., Tao, F., Hu, T., Anwer, N., Liu, A., Wei, Y., Wang, L. & Nee, A.Y.C. (2021). Enabling technologies and tools for digital twin. Journal of Manufacturing Systems, 58(B), 3-21. https://doi.org/10.1016/j.jmsy.2019.10.001 Rojek, I., & Mikołajewski, D. (2024). Use of Machine Learning to Improve Additive Manufacturing Processes. Applied Sciences. 14(15), 6730. https://doi.org/10.3390/app14156730 Syed, R., Suriadi, S., Adams, M., Bandara, W., Leemans, S.J.J., Ouyang, Ch., Hofstede, A.H.M., Weerd, I., Wynn, M.T., & Reijers, H.A. (2020). Robotic process automation: Contemporary themes and challenges. Computers in Industry, 115(103162). https://doi.org/10.1016/j.compind.2019.103162. Tao, F., Zhang, H., Liu, A., & Nee, A. Y. C. (2018). Digital twin in industry: State-of-the-art. IEEE Transactions on Industrial Informatics. 15(4), 2405-2415. https://doi.org/10.1109/TII.2018.2873186 Tikhonov, A. I., Sazonov, A. A., & Kuzmina-Merlino, I. (2022). Digital production and artificial intelligence in the aircraft industry. Russian Engineering Research, 42, 412–415. https://doi.org/10.3103/S1068798X22040293 Van Der Horn, E., & Mahadevan, S. (2021). Digital twin: Generalization, characterization and implementation. Decision Support Systems, 145(113524). https://doi.org/10.1016/j.dss.2021.113524. Wan, J., Li, X., Dai, H. N., Kusiak, A., Martínez-García, M., & Li, D. (2021). Artificial-intelligence-driven customized manufacturing factory: key technologies, applications, and challenges. Proceedings of the IEEE, 109(4), 377-398. https://doi.org/10.1109/JPROC.2020.3034808. Bhardwaj, V., Yadav, S., Kaur, N., & Anand, D. (2025). The Future of Work: Robotic Process Automation and its Role in Shaping Tomorrow’s Business Landscape. SN Computer Science, 6(111). https://doi.org/10.1007/s42979-02503668-0 Yin, J., Park, D., & Hong, S. (2023). Adoption barriers and future growth of smart wearable technologies in manufacturing. Journal of Industrial Engineering and Management, 15(3), 227-239. https://doi.org/10.3926/jiem.4441 Yu, W., Patros, P., Young, B., Klinac, E., & Walmsley, T.G. (2022). Energy digital twin technology for industrial energy management: Classification, challenges, and future. Renewable and Sustainable Energy Reviews, 161(112407). https://doi.org/10.1016/j.rser.2022.112407. Závadský, J., & Závadská, Z. (2018). Quality managers and their future technological expectations related to Industry 4.0. Total Quality Management & Business Excellence, 31(7–8), 717–741. https://doi.org/10.1080/14783363.2018.1444474 Zhang, C., Issa, H., Rozario, A., & Soegaard, J.S. (2023). Robotic process automation implementation case studies in accounting: A Beginning to End Perspective. Accounting Horizons, 37(1), 193–217. https://doi.org/10.2308/HORIZONS-2021-084 Zheng, Y., Yang, S., & Cheng, H. (2019). An application framework of digital twin and its case study. Journal of Ambient Intelligence and Humanized Computing, 10, 1141–1153. https://doi.org/10.1007/s12652-018-0911-3 147 DIGITALIZATION AS A KEY DRIVER OF ECONOMIC DEVELOPMENT IN THE MUSIC INDUSTRY IN EUROPEAN UNION Maria Bartekova, Sabina Janikovicova Bratislava University of Economics and Business Faculty of Business Management [email protected], [email protected] DOI: 10.35011/IDIMT-2025-147 Keywords Digitalization; Music Industry; Streaming; Sustainability Abstract Digitalization is fundamentally changing the way the music industry operates and creating new economic opportunities, particularly through the rise of streaming platforms. This article examines the relationship between countries’ economic development and music streaming revenues within the European Union. The aim is to determine whether countries with higher economic maturity are experiencing greater growth in digital music distribution. The research is based on data from the Statista database for the years 2015 to 2023 and uses correlation and regression analysis together with non-parametric tests (Kruskal-Wallis and Dwass-Steel-Critchlow-Fligner). The results show that music streaming revenues are statistically significantly higher in countries with higher economic development, while indicators such as real GDP growth rate and household spending did not show a consistent relationship. The findings highlight the existing digital divide between countries and highlight the need for targeted support for less developed economies in the areas of digital infrastructure and sustainable development. Digitalization can thus be not only an engine of economic growth, but also a means of ensuring the environmental and social sustainability of the music sector. 1. Introduction Digitalization is transforming the music industry in fundamental ways, altering how music is produced, distributed, and monetized (Baym et al., 2021; Brusila et al., 2022). This transformation is particularly relevant in the context of the European Union, where disparities in digital infrastructure and economic development may influence the extent to which countries can benefit from these changes. This study aims to understand the relationship between economic maturity and growth in music streaming revenues—a key indicator of the digital economy in the cultural sector. By exploring this relationship, we seek to inform policymakers, industry stakeholders, and scholars about the digital divide and its implications for sustainable development within the music industry. The music industry has undergone several fundamental changes, the most significant of which was digitalization. As Lozic & Cikovic (2023) state, the development of this industry was not only about 148 the invention of the phonograph, but mainly about the possibilities of mass production and distribution (Ruiz-Resto, 2023). The transition to digital formats was difficult - physical carriers replaced digital files and the Internet enabled their rapid dissemination (Denk et al., 2022). This trend led to widespread piracy, which significantly weakened the profits and position of record companies. A key turning point came in 1991 with the Napster platform, which enabled online music sharing. The aim of this research paper is to examine the relationship between the growth of music streaming revenues and the economic maturity of countries, assuming that countries with higher economic maturity experience stronger growth in revenues from digital distribution platforms. Digitalization, represented by the growing importance of streaming services, is seen as one of the main catalysts for the transformation of the music industry. 2. Literature review 2.1 Disruptive Innovation and the Music Industry The transition to digital formats disrupted the traditional structure of the music industry. Services like Napster (1991) and later iTunes revolutionized how music was consumed, initiating a shift from physical carriers to digital downloads and eventually streaming. iTunes, in particular, represented a key turning point by legitimizing digital music sales and providing a new economic model (RuizResto, 2023). This wave of disruption laid the foundation for the emergence of music-tech ecosystems and global platforms. While physical carriers are experiencing a decline in sales, digital streaming is experiencing a significant increase and has become the majority source of digital music distribution. However, this transition also brings new environmental challenges – streaming music requires enormous amounts of energy to power servers in data centres around the world Chalmers et al., 2021). In response, some digital platforms are starting to work with organizations that help them offset their carbon footprint by supporting reforestation projects or investing in renewable energy sources. An example is Spotify, which is switching from traditional data centres to the Google Cloud platform, which claims to operate using low-carbon energy sources (Chin et al., 2023; Pizzolitto, 2024). The traditional three-part segmentation of the music industry into publishing, recording and live music has been overcome. These traditional sectors have been joined by a fourth line of production and sales known as music-tech. This sector is particularly well developed in China and represents a significant shift in the structure of the global music industry (Zhang & Negus, 2021). In the process of digital transformation, traditional sectors of the music industry have a historic opportunity to connect with the new music-tech sector, thereby finding new paths of development and economic sustainability. A significant phenomenon of today is the emergence and development of export music agencies, which focus on presenting new musical talents at the international level and streamline the exchange of information between professionals in the music industry. Although digital distribution has significantly changed the functioning of the music industry, marketing and distribution alone are not sufficient means to ensure its functioning. Offline cooperation, which is ensured by export agencies, is equally important (Hosseini & Rajabipoor Meybodi, 2023). 2.2 Economic sustainability in the digital era Even though sales of physical music carriers are declining, we can currently observe a significant increase in concert attendance in the live music industry - an industry that includes live music festivals and concerts. Thanks to new ways of marketing and distributing advertising through digital platforms, live concerts have become increasingly popular, representing a significant source of revenue at a time 149 when physical sales are declining (Hagen, 2022). The music industry is also now dominated by global promotional monopolies and their music festivals, paid internet streaming services and YouTube. In addition, a number of music entrepreneurs are emerging, offering different methods of operating the music business and innovative product portfolios. Digitalization has also affected the way artists approach the release of music. Artists release a lot of music that is available for free on the internet but serves as the best promotion for their live performances. This strategy shows an adaptation to the new conditions of the digital market and represents a key aspect of economic sustainability in the current music industry (Brenner & Hartl, 2021). 3. Data and Methodology The aim of this research paper is to examine the relationship between the growth of music streaming revenues and the economic maturity of countries, assuming that countries with higher GDP per capita experience stronger growth in revenues from digital distribution platforms. Digitalization, represented by the growing importance of streaming services, is seen as one of the main catalysts for the transformation of the music industry. 3.1 Data source Empirical data were drawn from the Statista database, which provides reliable time series and international comparisons in the field of the music industry. The following indicators were used for the research purposes: • Annual music streaming revenues by country (2019,2020,2023), • Real GDP growth rate (USD, purchasing power parity), • Household Final Consumption Expenditure. The data were normalized and analysed using correlation and regression analysis to verify the relationship between the economic level of the country and the performance of streaming services. 3.2 Hypothesis formulation Based on the preliminary analysis and theoretical framework, the following hypotheses were established: • H₁: Higher economic development of the country (Real GDP growth rate) is positively correlated with the development of music streaming revenues. • H2: Higher economic development of the country (Household final consumption expenditure) is positively correlated with the development of music streaming revenues. The research conducted a descriptive statistical analysis of all variables to verify their distribution characteristics and ensure that the values fall within an acceptable range. Particular attention was paid to the skewness indicator as an indicator of distribution asymmetry, which may indicate deviations from normality. However, it is important to note that the standard error of the estimate in the t-test may underestimate the true error rate when comparing multiple groups, and therefore the t-test is not recommended for use when analyzing differences between more than two groups. In such cases, it is more appropriate to choose alternative methods such as analysis of variance (ANOVA). 150 4. Results and Discussion Figure 1 shows the global growth in the number of users of music streaming services from 2015 to 2023. The trend shows a continuous increase in the user base, indicating a significant shift in consumer preferences from physical media to digital content. The growth of users supports the economic sustainability of the music industry through increased subscription and advertising revenue. Figure 1. Music streaming – worldwide (number of users in mil.) Source: (Statista) Figure 2 illustrates the development of the average revenue per user of streaming services worldwide in USD. This indicator provides insight into the profitability of individual users and also highlights differences between countries according to the level of economic development. Higher average revenue often corresponds to higher purchasing power of users in economically developed countries. Figure 2. Average revenue per user (in USD) Source: (Statista) Kruskal-Wallis tests (Table 1-3) conducted for 2019, 2020 and 2023 showed that music streaming revenue (REV) showed consistently statistically significant differences between higher-income (High) and lower-income (Low) countries. The difference was particularly pronounced in 2019 (χ² = 8.7416, p = 0.0031, ε² = 0.3362), with this trend continuing in 2020 (χ² = 9.2569, p = 0.0023, ε² = 0.3560) and 2023 (χ² = 8.4524, p = 0.0036, ε² = 0.3251). These findings were also confirmed by pairwise comparisons, which showed significant differences in streaming revenue between country groups (p < 0.01 in all years). In contrast, household spending (HC) and gross domestic product growth (GDP_GR) did not show consistent statistical significance across the years studied. While GDP_GR showed a significant difference in 2019 (χ² = 5.1541, p = 0.0232), this result was not repeated in subsequent years, suggesting a weaker or less stable relationship. 0,0 200,0 400,0 600,0 800,0 1000,0 1200,0 1400,0 2017201820192020202120222023202420252026202720282029 Total 26,00 28,00 30,00 32,00 34,00 36,00 38,00 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 151 Table 1–3 reference Kruskal-Wallis test results for 2019, 2020, and 2023, confirming that streaming revenues (REV) are significantly higher in wealthier countries. In contrast, GDP growth (GDP_GR) and household consumption (HC) show inconsistent influence on music streaming revenues across the years studied. The digital divide is evident in the persistent gaps in streaming revenue. Higher-income countries benefit from better internet access, higher mobile penetration, and stronger digital consumer behaviour. The persistent revenue gap reflects structural inequalities, rather than temporary macroeconomic fluctuations. Table 1. One-Way Anova results - 2019 χ² df p ε² HC_19 2.0733 1 0.1499 0.0797 GDP_GR_19 5.1541 1 0.0232 0.1982 REV_19 8.7416 1 0.0031 0.3362 Pairwise comparisons - HC_19 W p High Low -2.0363 0.1500 Pairwise comparisons - GDP_GR_19 W p High Low 3.2107 0.0232 Pairwise comparisons - REV_19 W p High Low -4.1813 0.0031 Source: (own calculations according to Statista (2024) and Eurostat (2024)) Table 2. One-Way Anova results - 2020 χ² df p ε² HC 2.1529 1 0.1423 0.0828 GDP_GR 0.6242 1 0.4295 0.0240 REV 9.2569 1 0.0023 0.3560 Pairwise comparisons - HC W p High Low -2.0750 0.1424 Pairwise comparisons - GDP_GR W p High Low -1.1173 0.4297 Pairwise comparisons - REV W p High Low -4.3028 0.0023 Source: (own calculations according to Statista (2024) and Eurostat (2024)) Table 3. One-Way Anova results - 2023 Kruskal-Wallis χ² df p ε² HC_23 3.2822 1 0.0700 0.1262 GDP_GR_23 0.7836 1 0.3760 0.0301 REV_23 8.4524 1 0.0036 0.3251 Pairwise comparisons - HC_23 W p High Low -2.5621 0.0701 Pairwise comparisons - GDP_GR_23 W p High Low 1.2519 0.3761 Pairwise comparisons - REV_23 W p High Low -4.1115 0.0036 Source: (own calculations according to Statista (2024) and Eurostat (2024)) 152 These results indicate that the growth of music streaming revenue is more strongly associated with the overall economic level of a country than with annual fluctuations in GDP growth or household consumption. The findings of this study support the hypothesis that countries with higher economic maturity experience stronger growth in music streaming revenues. This result supports a broader literature that highlights that digital infrastructure, purchasing power, and established digital consumer behaviour are key factors for success in the digital content industry (Chin et al., 2023). Wealthier countries tend to benefit from better access to stable internet, higher smartphone penetration, and a technologically savvy population, all of which facilitate the use of subscription-based or advertising-based streaming services (Khlystova & Kalyuzhnova, 2023). While GDP growth and household final consumption expenditure were expected to have an impact on streaming performance, their inconsistent statistical significance suggests that they are less predictive than overall economic maturity. This is consistent with existing studies that emphasize the importance of long-term structural factors over short-term macroeconomic changes in digital market development (Brenner & Hartl, 2021). The persistent gap between highand low-income countries in streaming revenues over all years points to a persistent digital divide. This not only limits the economic sustainability of the music industry in less developed countries, but also makes it more difficult to gain equal access to global music markets (Geurts & Cepa, 2023). Moreover, it confirms previous findings that music-technology ecosystems thrive in more economically developed countries where innovation, investment, and regulatory support converge (Zhang & Negus, 2021). These findings call for targeted policy measures, including subsidies for digital infrastructure development, international cooperation, and capacity building in lower-income countries, to bridge the gap. The integration of sustainable practices, such as carbon-neutral streaming operations (Chalmers et al., 2021), should be globally available to ensure that digitalization contributes equitably to both the economic and environmental sustainability of the music sector. 5. Conclusion This paper demonstrates a strong and consistent relationship between a country’s economic development and the growth of music streaming revenues, with high-income countries having a significant advantage. While short-term economic indicators such as GDP growth and household consumption expenditure showed limited impact, the results reinforce the idea that broader economic maturity – reflected in stable infrastructure, consumer habits and digital readiness – is a key driver of success in the digital music economy. From a practical perspective, the research highlights both the opportunities and challenges of digitalization. While its benefits for economic sustainability are evident in developed countries, barriers such as poor infrastructure and lower purchasing power persist in less developed regions. This deepens inequalities in the global music industry. To mitigate these, it is necessary to invest in digital infrastructure, promote international cooperation and make the benefits of digitalization available to all regions. At the same time, music platforms should support fairer remuneration and better access for creators around the world. As the digital music ecosystem continues to evolve, ensuring both economic and environmental sustainability will require coordinated efforts across the public and private sectors. Only in this way SOCIAL MEDIA AND AI: CONTRIBUTOR, SELECTOR, …? 257 AI INNOVATIONS - THE DOUBLE-EDGED SWORD: EXPLORING POSITIVE AND NEGATIVE IMPLICATIONS OF ARTIFICIAL INTELLIGENCE IN SOCIAL MEDIA Antonín Pavlíček Prague University of Economics and Business Faculty of Informatics and Statistics [email protected] DOI: 10.35011/IDIMT-2025-257 Keywords Innovations; social media; AI; implications Abstract This paper investigates the multifaceted implications of Artificial Intelligence (AI) integration within social media platforms. Employing an analytical approach, it identifies 10 positive effects of AI in social media such as: enhanced content personalization, efficient fake news detection, improved user engagement, automated content moderation, advanced trend analysis, targeted advertising efficiency, real-time crisis management,increased accessibility, improved community building, enhanced data-driven insights and 10 negative effects, such as: amplification of bias, privacy concerns, creation of echo chambers, reduced transparency, user manipulation, risk of automation errors, increased polarization, algorithmic accountability issues, ethical dilemmas in moderation, and economic impacts on content creators. By synthesizing current scholarly literature, this study underscores the urgency of balanced governance to mitigate adverse effects while promoting responsible AI utilization in social media. 1. Introduction Artificial Intelligence (AI) has become a transformative force across numerous industries, with social media platforms being a prime example of its impactful integration. The advent of sophisticated AI technologies has significantly reshaped how users interact, share information, and engage with content online (Vinuesa et al., 2020). This dramatic shift reflects a broader technological evolution characterized by personalization, automated moderation, and intelligent content recommendation systems. Social media platforms such as Facebook, Instagram, Twitter, and TikTok have rapidly adopted AIdriven algorithms to optimize user experiences. These platforms leverage AI to analyze vast amounts of user-generated data, allowing them to curate highly personalized content feeds and recommendations. This targeted personalization has led to improved user satisfaction and increased platform engagement, establishing a new norm in digital user experiences (Schwartz et al., 2022). 258 One of the critical positive outcomes of AI's integration into social media is its ability to manage the vast scale and complexity of content. AI systems excel at tasks such as identifying trending topics, filtering out harmful or inappropriate content, and enhancing user engagement by providing relevant suggestions. The automation of these processes significantly improves operational efficiency and user experience, highlighting AI’s beneficial aspects within the social media ecosystem. Moreover, AI has become instrumental in addressing misinformation and disinformation, critical challenges facing modern societies. Algorithms designed to detect and flag fake news have increasingly been employed by major platforms, providing a frontline defense against information manipulation. AI’s capability to analyze linguistic patterns, cross-reference facts, and flag suspicious content contributes significantly to maintaining the integrity and reliability of information disseminated through social networks (Akhtar et al., 2023). Despite these advancements, AI's deployment in social media raises substantial ethical questions. The reliance on automated decision-making processes introduces potential biases inherent in training data and algorithmic design. AI systems trained on biased data can inadvertently perpetuate stereotypes and exacerbate inequalities, impacting marginalized communities disproportionately. Such biases underscore the necessity of rigorous ethical standards and transparency in AI development and deployment. The issue of transparency in algorithmic processes remains another significant concern. Users often lack insight into how AI-driven recommendations and content filtering mechanisms operate, creating a sense of opacity and mistrust. This lack of transparency complicates accountability and reduces user autonomy, as decisions influencing their online experiences are made without clear explanation or recourse (Schwartz et al., 2022). Privacy implications also represent a critical area of concern. Social media platforms collect extensive user data, often processed by complex AI systems to generate insights and drive targeted advertising. While personalized content enhances user experiences, it simultaneously elevates privacy risks, as sensitive information can be mishandled or exploited. Increased reliance on AI systems necessitates robust data protection mechanisms and enhanced user control over personal data (Tucker, 2019). AI-driven personalization algorithms can create "echo chambers" or "filter bubbles," wherein users are predominantly exposed to content reinforcing their existing beliefs and biases. Such environments contribute to polarization, undermining the diversity and quality of public discourse. The long-term societal impact of echo chambers on democratic processes and social cohesion warrants serious consideration. There is an urgent need for robust regulatory frameworks and governance structures. Policymakers, technologists, and ethicists must collaborate to develop effective oversight mechanisms to mitigate the adverse effects of AI. Emphasis should be placed on ensuring transparency, accountability, and ethical AI use, addressing both immediate and long-term challenges associated with its implementation (Novák et al., 2025). The integration of AI in social media presents both significant opportunities and substantial risks. Balancing these opposing forces requires a comprehensive understanding of AI’s capabilities and limitations, informed by ongoing scholarly research and practical insights. This paper aims to provide such understanding, setting the stage for a detailed examination of AI's dual implications in social media. 259 2. Methods A qualitative literature review methodology was utilized, examining peer-reviewed journal articles, conference proceedings, and authoritative reports published between 2018 and 2024. Databases such as IEEE Xplore, Web of Science, and Scopus were systematically searched using keywords such as "AI ethics in social media," "AI privacy," "fake news detection," and "AI-driven recommendations." Selected studies underwent thematic analysis to identify recurrent patterns and critical issues within the research domain. 3. Results 3.1 Positive Effects of AI in Social Media AI significantly enhances the user experience by providing tailored content recommendations, improving user engagement metrics substantially (Schwartz et al., 2022). Additionally, AI has proven effective in detecting misinformation and fake news, thus helping maintain informational integrity on platforms (Akhtar et al., 2023). List of Major Positive Effects of AI in Social Media 1. Enhanced Content Personalization 2. Efficient Fake News Detection 3. Improved User Engagement 4. Automated Content Moderation 5. Advanced Trend Analysis 6. Targeted Advertising Efficiency 7. Real-time Crisis Management 8. Increased Accessibility 9. Improved Community Building 10. Enhanced Data-Driven Insights Enhanced Content Personalization AI algorithms analyze extensive user behavior and preferences data, tailoring content specifically to individual interests. This personalization significantly enhances user experiences, increasing satisfaction and platform engagement (Schwartz et al., 2022). Users receive highly relevant content that aligns closely with their tastes and needs. Personalization also reduces information overload by streamlining content delivery, making the user experience more manageable and enjoyable. It transforms user interaction patterns by providing seamless and intuitive navigation through massive content streams. Moreover, tailored content encourages users to spend more time on platforms, creating deeper userplatform relationships that benefit both parties, promoting sustained usage and loyalty. 260 Efficient Fake News Detection AI-driven systems play a critical role in the identification and management of misinformation. Machine learning algorithms detect patterns and inconsistencies indicative of false information, significantly enhancing platform reliability (Akhtar et al., 2023). This capability has become vital in combating misinformation, especially during critical public events. AI also supports real-time intervention, enabling platforms to swiftly flag and remove misinformation. This rapid response is essential in limiting the viral spread of harmful content and maintaining the integrity of public discourse. Additionally, continuous algorithmic improvements increase the accuracy and efficiency of fake news detection systems, making them increasingly robust against evolving misinformation tactics. Improved User Engagement AI algorithms enable social media platforms to boost user interaction by predicting and delivering engaging content. User engagement metrics, such as likes, comments, and shares, have improved significantly due to AI's ability to identify content that resonates with specific user demographics (Schwartz et al., 2022). This predictive ability helps maintain user interest and loyalty, reducing churn rates and enhancing overall platform stability and growth, especially with chatbot usage (Potančok & Radváková, 2023) Furthermore, enhanced user engagement supports better monetization opportunities for platforms and content creators alike, creating a vibrant digital economy (Sudzina, 2013). Automated Content Moderation AI enables efficient and consistent moderation of massive content volumes. Automated moderation reduces the burden on human moderators, enhancing operational efficiency and maintaining higher content quality standards. AI-driven moderation systems quickly identify and address inappropriate or harmful content, ensuring compliance with platform guidelines and regulatory requirements. Automated moderation also supports scalability, allowing platforms to manage content effectively even as user bases and content volumes expand rapidly. Advanced Trend Analysis AI enhances the detection and analysis of emerging trends by examining vast amounts of data across social media platforms. This capability allows platforms to rapidly identify and leverage trending topics to maximize engagement and relevance. Trend analysis helps platforms proactively manage user content strategies, providing valuable insights for marketers, content creators, and users. Targeted Advertising Efficiency AI greatly enhances the efficiency and accuracy of targeted advertising on social media platforms by analyzing user data to understand preferences, buying behaviors, and interests. These insights enable precise targeting, significantly improving the effectiveness and relevance of advertisements shown to users. [Document text truncated for crawler view.]