Socio-technical Perspectives of Conversational Systems Design

Socio-technical Perspectives of Conversational Systems Design Investigating Human-AI Interaction beyond Speed, Accuracy and Task Success Prof. Dr. Stephan Schlögl | CRYSTAL-RTTH Fall School, 12 November 2025, Bilbao, Spain Source: AI generated by Google Gemini 1.5 Pro Input prompt: “create an image showing an artificial intelligence for a talk entitled: Acceptance, Trust, Personality & Gender Investigating Inherent Challenges of Human-AI Interaction” Source: AI generated by Google Gemini 2.5 Flash Input prompt: “An intelligent machine. Put it into a business setting.” ABOUT ME… Prof. Dr. Stephan Schlögl Professor of Human-Centered Computing email: [email protected] phone: +43 512 2070 – 3535 Research interests - Human-Computer Interaction -Interactive Systems - Natural Language Processing -Computer Supported Cooperative Work -Information & Knowledge Management Source: AI generated by Google Gemini 1.5 Pro Input prompt: “create an image showing an artificial intelligence for a talk entitled: Acceptance, Trust, Personality & Gender Investigating Inherent Challenges of Human-AI Interaction” Source: AI generated by Google Gemini 2.5 Flash Input prompt: “An intelligent machine. Put it into a business setting.” 29 Bachelor and Master Programs + Executive Programs (MBA, MSc, LL.M.) Source: AI generated by Google Gemini 1.5 Pro Input prompt: “create an image showing an artificial intelligence for a talk entitled: Acceptance, Trust, Personality & Gender Investigating Inherent Challenges of Human-AI Interaction” Source: AI generated by Google Gemini 2.5 Flash Input prompt: “An intelligent machine. Put it into a business setting.” Further info: http://www.mci.edu Let‘s talk about CUIs… 1950s… TURING TEST Alan Turing, 1950 Credits: By Juan Alberto Sánchez Margallo (Diseño propio) [CC BY-SA 2.5 (https://creativecommons.org/licenses/by-sa/2.5)], via Wikimedia Commons “Is it a human or is it a machine?” 1960s… 1966 Joseph Weizenbaum, MIT Source: https://www.derstandart.at Source: https://www.derstandart.at Source: https://www.techradar.com/computing/artificial-intelligence/chatgpt-is-getting-smarter-but-its-hallucinations-are-spiraling The problem of language… Source: https://pixabay.com/photos/wildlife-ape-gorilla-4328243/ by: garten-gg | Pixabay License LANGUAGE Language as a human capacity… -Approx. 100,000 years ago we learned how to speak -Approx. 7,000 years ago we learned how to write -Although, there are other attributes that are uniquely human (e.g. wearing cloth, art, etc.) it was language Alan Turing based his test on! Source: https://pixabay.com/en/binary-one-cyborg-cybernetics-1536617/ by: Gerald | CC0 Creative Commons HUMAN LANGUAGE Natural languages... -Natural languages (such as English, or Spanish, etc.) cannot be categorized as a definitive set of sentences • “Not to be invited is sad” vs. “To be not invited is sad” • Hence, we may define a natural language model as a probability distribution over sentences -Also, natural language is ambiguous •“He saw her duck” Source: https://pixabay.com/en/speak-talk-microphone-tin-can-can-238488/ By: RyanMcGuire | CC0 Creative Commons COMPUTER LANGUAGE Formal languages... -Formal languages (such as programming languages) have precisely defined language models •“print(2+2)” is a legal program in Python whereas “2)+(2 print” is not • Hence, a language is specified by a set of rules i.e. a grammar - In addition, formal languages have rules that define the meaning or semantics - E.g. the “meaning” of “2 +2” is 4 and “1/0” is an error Source: https://pixabay.com/en/programming-computer-environment-1857236/ by: 3844328 | CC0 Creative Commons So, how can we figure out what people think about the tech we build? https://pixabay.com/photos/man-confused-young-male-glasses-5914349/ by: Usman Yusaf | Pixabay License Determinants of Technology Use (AI, CUI, VUI, etc.) Source: https://pixabay.com/photos/ai-robot-artificial-intelligence-7977960/ by: Alexandra Koch | Pixabay License TECHNOLOGY ACCEPTANCE Technology Acceptance Model (TAM 1) Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1989). Technology acceptance model. J Manag Sci, 35(8), 982-1003. Those two constructs tend to explain between 40% and 70% of the variance in actually using a technology Based on Ajzen and Fishbein‘s Theory of Reasoned Action (TRA) and Ajzen’s successive Theory of Planned Behavior (TPB) •Fishbein, M., & Ajzen, I. (1980). Predicting and understanding consumer behavior: Attitude-behavior correspondence. Understanding attitudes and predicting social behavior, 1(1), 148-172. •Ajzen, I. (1985). From intentions to actions: A theory of planned behavior. In Action control: From cognition to behavior (pp. 11-39). Berlin, Heidelberg: Springer Berlin Heidelberg. TECHNOLOGY ACCEPTANCE Affinity for Technology Scale (ATI) Franke, T., Attig, C., & Wessel, D. (2019). A personal resource for technology interaction: development and validation of the affinity for technology interaction (ATI) scale. International Journal of Human–Computer Interaction, 35(6), 456-467. •Affinity for Technology •To what extent do people “like” technology? TECHNOLOGY ACCEPTANCE Virtual Agents (VAAQ), Robot (RAQ), and Synthetic Virtual Agent Voice Acceptance (VAVAQ) Esposito, A., Amorese, T., Cuciniello, M., Esposito, A. M., & Cordasco, G. (2023). Do you like me? Behavioral and physical features for socially and emotionally engaging interactive systems. Frontiers in Computer Science, 5, 1138501. •What about the acceptance of our chatbots, virtual agents and robots? This also includes measures pragmatic qualities (practical, result-oriented), hedonic qualities (pleasure, fun, etc.) and perceived attractiveness! TRUST IN TECHNOLOGY Components and Measures of Trust in Technology Mcknight, D. H., Carter, M., Thatcher, J. B., & Clay, P. F. (2011). Trust in a specific technology: An investigation of its components and measures. ACM Transactions on management information systems (TMIS), 2(2), 1-25. The general tendency to be willing to depend on technology across a broad spectrum of situations and technologies. The belief that success is likely because the situation is normal, favorable, or well-ordered. The belief that success is likely because contextual conditions like promises, contracts, regulations and guarantees are in place. Reflects beliefs that a specific technology has the attributes necessary to perform as expected in a given situation in which negative consequences are possible. Assumption that technologies are usually consistent, reliable, functional, and provide the help needed Assumption that the use of technology leads to a better outcome. Source: https://pixabay.com/vectors/einstein-albert-physics-quantum-8041625/ by: Mohamed Hassan | Pixabay License ꟷYet another problem of intelligence ꟷ SOCIAL INTELLIGENCE Situational Awareness Presence Authenticity Clarity Empathy Source: https://pixabay.com/en/binary-code-privacy-policy-woman-2175285/ by: geralt | CC0 Creative Commons Social Intelligence Albrecht, K. (2009). Social intelligence: The new science of success. John Wiley & Sons. Some studies… Summary & Key Findings: •Interview Study with n=12 experts from different fields who had contributed to the development of AI agents •We identified 5 characteristics of agent authenticity (4) be coherent (5) anthropomorphize (3) be transparent i.e., have a purpose (2) learn from experience (1) show strong conversational behavior 2016 Neururer, M., Schlögl, S., Brinkschulte, L., & Groth, A. (2018). Perceptions on Authenticity in Chat Bots. Multimodal Technologies and Interaction, 2018, 2(3). doi: 10.3390/mti2030060 Summary & Key Findings: •Delphi Study with n=21 multi-disciplinary experts from academia (n=11) and industry (n=10) •14 distinct characteristics socially intelligent conversational agents should offer 1. Context-related Action 2. Reflective Language 3. Enculturation 4. Customizability 5. Engagement 6. Consistency 7. Depth 8. Continuous Interaction 9. Respectful Honesty 10. Justifiability 11. Establish/Maintain Relationships 12. Respectful Acting 13. Otherness 14. Individual Personality Albrecht, K. Social Intelligence: The New Science of Success; Jossey-Bass: New York, NY, USA, 2006. 2018 Brinkschulte, L., Schlögl, S., Monz, A., Schöttle, P., & Janetschek, M. (2022). Perspectives on Socially Intelligent Conversational Agents. Multimodal Technologies and Interaction, 6(8). doi: 10.3390/mti6080062 Summary & Key Findings: •Online survey (n=367) based on the Big 5 personality traits and McKnights model of trust in technology Big Five: Tupes, E.C.; Christal, R.E. Recurrent personality factors based on trait ratings. J. Personal. 1992, 60, 225–251 Trust in Technology: McKnight, D.H.; Carter, M.; Thatcher, J.B.; Clay, P.F. Trust in a specific technology: An investigation of its components and measures. ACM Trans. Manag. Inf. Syst. (TMIS) 2011, 2, 1–25. •Personality does NOT affect people’s propensity to trust •Affinity for Technology does NOT affect people’s trusting believes in intelligent virtual assistants •Propensity to trust in general technology positively correlates with trusting believes in intelligent virtual agents 2019 Schadelbauer, L., Schlögl, S., & Groth, A. (2023). Linking Personality and Trust in Intelligent Virtual Assistants. Multimodal Technologies and Interaction, 7(6). doi: 10.3390/mti7060054 Summary & Key Findings: •Online survey (n=188) focusing on the acceptance of Social Assistance Robots in a German-speaking population and how it is related to people’s personality Measures •Technology Acceptance based on the Almere Model •Personality based on the short version of the Big Five Inventory extended by resilience •Technology Experience •Expectations Results •Findings suggest that personality plays a significant role in the acceptance of SAR technologies •We found significant correlations between age, gender, education, personality, resilience, experience, expectations, and technology acceptance and its subdimensions. •In particular, agreeableness (kind, cooperative, empathetic) was found to positively correlate with acceptance •On the other hand, neuroticism (negative emotions like anxiety, anger, and depression) was found to negatively correlate with acceptance. 2019 Gessl, A. S., Schlögl, S., & Mevenkamp, N. (2019). On the perceptions and acceptance of artificially intelligent robotics and the psychology of the future elderly. Behaviour & Information Technology, 38(11), 1068-1087. Some current studies… What is the impact of XAI explanations on users’ trust miscalibration? (under review) How does the use of GenAI affect knowledge workers’ perceptions regarding the quality of and responsibility for their work? (research in progress) What is the role of AI in complex decision making processes of start-ups? (research in progress) Source: AI generated by Google Gemini 1.5 Pro Input prompt: “generate an image of an ai that thinks” Prof. Dr. Stephan Schlögl MCI | The Entrepreneurial School Universitätsstraße 15 6020 Innsbruck, AUSTRIA Email: stephan.schloe[email protected] Phone: +43 512 2070 3535 linkedin.com/in/stephanschloegl/ Source: AI generated by Google Gemini 1.5 Pro Input prompt: “create an image of an ai saying thank you”