D3.6 - Overview over ELSA services & description of their purpose and function for AI development in agri-food

D3.6 OVERVIEW OVER ELSA SERVICES & DESCRIPTION OF THEIR PURPOSE AND FUNCTION FOR AI DEVELOPMENT IN AGRIFOOD [31.12.2024] Ref. Ares(2024)9286201 - 31/12/2024 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 2 Project cofunded by the European Commission within the Digital Europe Programme Dissemination Level PU Public □ CO Confidential, only for members of the consortium (including the Commission Services) x CL Classified, as referred to in Commission decision 2001/844/EC □ Deliverable number: D3.6 Deliverable name: Overview over ELSA services & description of their purpose and function for AI development in agri-food Deliverable Description: Document presenting the ELSA services offered by the TEF, including details of the purposes and expected impacts of these services Work package: WP3 Authors Can Atik (WR), Mireille van Hilten (WR), Roberto García (UdL), Rosa Gil (UdL), Georgios Tsagdis (WU), Mehran Rad (RISE), Anne Bruinsma (WR) (Acknowledgement: Krzysztof Sieczkarek (L-PIT)) Reviewer(s) Ankur Mehtani (LNE) Lead WP: Laboratoire national de métrologie et d'essais (LNE) Lead task: Wageningen Research (WR) Document Revision History Date Issue Author/Editor/Contributor Summary of the main changes 14/02/2024 V1 Can Atik (WR) First draft version. Structure and key information in place. 03/04/2024 V2 Georgios Tsagdis (WUR) Development of flow, argumentation, and presentation. 25/06/2024 V3 Can Atik (WR) First draft version of the deliverable in clear format. 12/11/2024 V4 Can Atik (WR) The final draft version of the deliverable in clear format. 12/12/2024 V5 Ankur Mahtani (LNE) Internal review of the deliverable 27/12/2024 V6 Can Atik (WR) The final version to be submitted 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 3 Table of Contents Abstract ............................................................................................................................................................................. 4 Executive Summary ........................................................................................................................................................... 5 1. Introduction............................................................................................................................................................... 7 2. Methodology of the Deliverable ................................................................................................................................ 9 2.1 Exploring the Current State of ELSA Research (Literature Review) ......................................................................... 9 2.2 Listing and Analyzing ELSA and LCA Services in the AgrifoodTEF Project ................................................................ 9 3. Background and State-of-the-Art in Ethical, Legal, and Social Aspects (ELSA) Research & Relationship with Life Cycle Assessment (LCA) ................................................................................................................................................... 10 3.1 Overview of ELSA Approach .................................................................................................................................. 10 3.2 Background and Conceptual Framework .............................................................................................................. 10 3.3 Responsible Research and Innovation (RRI) and ELSA Research ........................................................................... 13 3.4 Value Sensitive Design (VSD) and ELSA research ................................................................................................... 15 3.5 ELSA & AI Use in Agriculture .................................................................................................................................. 17 3.5.1 Ethical Aspects of AI Technology in Agriculture .............................................................................................. 18 3.5.2 Legal Aspects of AI Technology in Agriculture ................................................................................................ 19 3.5.3 Social Aspects of AI Technology in Agriculture ............................................................................................... 22 3.6 ELSA Labs and Practice in the Agri-food Sector ..................................................................................................... 25 3.6.1 ELSA Labs ........................................................................................................................................................ 25 3.6.2 Application under the AgrifoodTEF Project .................................................................................................... 28 3.7 LCA to Complement ELSA in the Agrifood Sector .................................................................................................. 31 3.7.1 Introduction .................................................................................................................................................... 31 3.7.2 Life Cycle Assessment in Brief ........................................................................................................................ 32 3.7.3 LCA in Agrifood Tech ...................................................................................................................................... 33 3.7.4 LCA to complement ELSA ............................................................................................................................... 35 4. ELSA and LCA Services Offered within the AgrifoodTEF .............................................................................................. 37 4.1 List of ELSA services ............................................................................................................................................... 37 4.2 List of LCA Services ................................................................................................................................................ 39 5. Purpose and Expected Impacts of ELSA & LCA Services under AgrifoodTEF ............................................................... 40 5.1 Purposes and Expected Impacts of the ELSA Services ........................................................................................... 40 5.2 Purposes and Expected Impacts of the LCA Services ............................................................................................. 46 6. Conclusion ................................................................................................................................................................... 48 References ....................................................................................................................................................................... 51 Acknowledgements to co-funding agencies .................................................................................................................... 58 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 4 Abstract The AgrifoodTEF project offers Ethical, Legal, and Social Aspects (ELSA) and Life Cycle Assessment (LCA) services in order to support the transition from innovative AI solutions to successful market applications in the agri-food sector. This report explores the current ELSA research and demonstrates how LCA may complement ELSA to promote the responsible development and deployment of AI technologies. It explores the ELSA and LCA services under the AgrifoodTEF service catalogue and outlines the purpose and impact of these services. ELSA services help address ethical, legal, and social challenges while ensuring compliance with regulations and fostering human-centred design. LCA services evaluate and optimise the environmental impact of technologies, supporting sustainability goals. Together, these services have the potential to enable a comprehensive approach to responsible innovation by evaluating ethical, legal, and societal aspects as well as environmental considerations. The report also highlights the potential for developing more comprehensive ELSA services that consider ethical, legal and social aspects more in-depth and with quadruple helix stakeholder workshops to enhance collaboration and inclusivity. Additionally, raising awareness about these services, particularly in underutilised regions, is essential for broader adoption and impact. By aligning AI innovations with ethical standards, legal rules, societal values, and sustainability priorities, the ELSA and LCA services are expected to help sectoral stakeholders better understand the ethical, legal, social and environmental risks and opportunities. Thus, this report highlights the ELSA and LCA services’ importance in shaping responsible innovation and contributing to a sustainable digital transformation of the agri-food ecosystem. Future deliverables, such as D3.7, will build on these findings to evaluate the performance of these services and provide recommendations for continuous improvement. 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 5 Executive Summary The AgrifoodTEF project aims to foster sustainable and efficient food production by empowering AI innovators with the necessary validation tools to bridge the gap between innovative ideas and successful market products. In this regard, the AgrifoodTEF initiative brings together inter alia advanced Ethical, Legal, and Social Aspects (ELSA) and Life Cycle Assessment (LCA) services to support the responsible development and deployment of technologies in the agri-food sector. This deliverable explores the state of the art in ELSA research and how LCA may be a useful complement to the ELSA framework for the agri-food domain. In particular, the deliverable provides an overview of the ELSA and LCA services available under the AgrifoodTEF service catalogue 1 , detailing their purposes and expected impacts. These services are designed to assess, guide, and validate AI and robotics solutions in real-world conditions. The integration of these services is crucial for promoting ethical AI development, ensuring regulatory compliance, and fostering social acceptance in addition to environmental considerations in the agri-food sector. The ELSA and LCA services can help stakeholders mitigate risks and align their AI-driven solutions with required standards and obligations. By using these services, stakeholders are supported in making informed decisions about technology investments, optimizing their products, and aligning their operations with necessary requirements. Such a comprehensive framework helps build trust among companies, policymakers, and end-users, creating a robust foundation for the seamless adoption of transformative technologies. The findings in this report highlight the critical role of these services in facilitating the responsible and sustainable adoption of AI technologies in the agri-food sector. Key insights include: • Purpose and Impact: ELSA services aim to empower stakeholders to identify and address ethical, legal, and societal challenges, such as fairness, transparency, sustainability, autonomy, and labour as well as alignment with legal frameworks and human-centric design principles. LCA services complement this by enabling stakeholders to assess and optimise the environmental performance of their technologies, ensuring alignment with sustainability goals. • Synergy Between ELSA and LCA: The ELSA and LCA services offer a unique potential for addressing ethical, legal and societal aspects, and environmental challenges in a comprehensive way as LCA can assess environmental considerations in detail after ELSA aspects have been identified. • Future Directions: The analysis revealed that while existing services emphasise legal compliance and ethical considerations, the inclusion of stakeholder workshops remains an opportunity for growth, especially considering the emphasis in the literature to include quadruple helix stakeholders (researchers, businesses, government, and citizens) in ELSA practice. Although currently not provided as a service in AgrifoodTEF, such workshops could foster collaboration among stakeholders enhancing the participatory and inclusive nature of ELSA considerations. 1 To validate, test or evaluate agrifood AI solutions, this catalogue helps potential customers to find related services: https://www.agrifoodtef.eu/catalogue-of-services 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 6 • Awareness and Adoption: The adoption of ELSA and LCA services varies across regions, with some services yet to be utilized by potential AgrifoodTEF customers. Targeted efforts to raise awareness about the value of these services, particularly in regions with recent participation in the project, will be pivotal in increasing engagement and impact. In conclusion, the ELSA and LCA services under AgrifoodTEF align technological advancements with ethical considerations, regulatory requirements, societal values, and environmental priorities, and promote responsible innovation, aim to enhance trust among stakeholders and support sustainability. They provide the tools and frameworks necessary for stakeholders to navigate complex challenges, seize new opportunities, and contribute to a more inclusive, sustainable, and resilient agrifood ecosystem through AI innovation. This report highlights the transformative potential of ELSA and LCA services in fostering AI innovation by complying with ethical and legal standards as well as safeguarding societal and environmental interests. The findings of this report also pave the way for further refinement of these services, with the upcoming deliverable, D3.7, set to provide performance evaluations and insights for continuous improvement. Ultimately, these services establish a benchmark for responsible innovation, empowering stakeholders to address complex challenges, seize new opportunities, and contribute to a more sustainable and resilient agrifood ecosystem. 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 7 1. Introduction AgrifoodTEF is a European network of physical and digital facilities dedicated to supporting innovators in the agrifood sector. Its mission is to provide tailored services that assess and validate AI and robotics solutions under real-world conditions, thereby maximising their impact on agriculture. The network offers a comprehensive catalogue of services across various sectors, including arable farming, horticulture, tree crops, viticulture, greenhouses, livestock farming, and food processing. These services are customisable to meet specific customer needs, ensuring an efficient process. The AgrifoodTEF adopts the metaphor of a shop as a business to conceptualise its approach. From the very beginning of the project, the AgrifoodTEF is envisioned as being fully operational and delivering testing and experimentation services. The services offered by the AgrifoodTEF can be categorised into three main types: 1. Physical Testing and Experimental Environments related to WP1 These are designed to directly support customers, including SMEs, start-ups, and larger companies that develop products for farmers. They provide the physical infrastructure necessary for testing and experimentation activities. 2. Digital Testing and Experimental Environments related to WP2 To accommodate the testing of software solutions and AI models, a digital environment is essential. This environment supports simulation, modelling, and visualization activities. The integration of digital and physical environments is expected to create a seamless testing experience, with the digital component offering additional opportunities for international market expansion. 3. Ethical, Legal, Social, Economic, Business, Conformity, and Certification Services related to WP3 The "shop" also delivers services addressing these critical aspects, ensuring a comprehensive approach to support stakeholders in the agri-food sector. These three types of services form the foundation of AgrifoodTEF are organised into distinct work packages (WP1, WP2, and WP3), which represent the project's core pillars. This deliverable is under the WP3, conformity and Ethical, Legal, and Social Aspects (ELSA) testing infrastructure. The WP3 also covers services for conformity, safety, cybersecurity, Life Cycle Assessment (LCA), and regulatory sandboxes: “The general goal of this WP is to set up and implement non-functional and digital services that support the development and the market introduction of ‘standard compliant’ and trustworthy AI and AI-powered robotics technologies. Specific attention will also be given to AI solutions related to environmental sustainability and animal welfare in current and upcoming ethical and legal regulation, labelling, certification, and compliance.” All the services under the service catalogue are available at the AgrifoodTEF website. 2 2 https://www.agrifoodtef.eu/catalogue-of-services 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 8 This deliverable primarily focuses on ELSA services that have been added to the service catalogue of AgrifoodTEF considering the clear title of the D3.6: “Overview over ELSA services & description of their purpose and function for AI development in agri-food” & definition of D3.6: “Document presenting the ELSA services offered by the TEF, including details of the purposes and expected impacts of these services”. This deliverable also explores Life Cycle Assessment (LCA) services as an additional component considering the Task 3.2 title: “Collect, improve and develop ELSA and LCA services” and related descriptions of the task. Other deliverables within WP3 provide separate detailed analysis of the remaining important elements of the WP3. In particular, this deliverable explores the ELSA literature, provides the official descriptions of the ELSA and LCA services under the AgrifoodTEF service catalogue, and discusses these services' purposes and expected impacts. This report, therefore, represents the state-of-the-art ELSA research and the current status of the ELSA and LCA services under the AgrifoodTEF service catalogue and provides a discussion on the purposes and expected impacts of these services. Due to the dynamic nature of the service development within the AgrifoodTEF project 3 , the services investigated in this report are limited to the ones, which have already been developed and registered by December 2024. In this regard, this deliverable aims to answer the following questions; -i. What are the origins, evolution and current state of ELSA research? What is the scope of legal, ethical and social assessments in the ELSA research? How are ELSA and LCA related in the context of the agri-food sector? -ii. What are the ELSA services and LCA services offered within the scope of the AgrifoodTEF project? What are the purposes and expected impacts of these services for the agri-food systems? This investigation starts with establishing ELSA research origins, evolution, and present scope. The main objective is to establish the necessary foundation regarding evolution and the state-of-the-art in ELSA research. This is the starting point to examine how this research framework can support the development and testing of AI solutions in the agri-food sector, which is the unique connection with the purpose and mission of the AgrifoodTEF project, in general, and its ELSA services, in particular. Thus, the deliverable can delve into how ELSA practice is informed by a contextual understanding of ELSA considerations on AI technologies in the sector by particularly focussing on the available services under the AgrifoodTEF project service catalogue and their purpose and expected impacts in agri-food systems. The structure of the deliverable is designed to comprehensively explore and analyse the ELSA framework within the context of AI applications in the agrifood sector and given services under the AgrifoodTEF project. The deliverable is organized into six main sections. Section 2 briefly explains the methodology of the deliverable to address the research questions. Section 3 provides an extensive overview of ELSA research and conceptual framework. It covers the historical context and evolution of ELSA research, its relationship with other 3 New services will be added to the catalogue to further expand the AgrifoodTEF offering, aiming to address additional ELSA and LCA aspects and ensure comprehensive coverage of emerging needs in the agri-food sector. 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 9 frameworks such as Responsible Research and Innovation (RRI) or Value Sensitive Design (VSD), and the current state of ELSA research. The section also examines ELSA lab practices, particularly in the agri-food sector, and the relationship of ELSA with LCA in the agri-food context. Section 4 lists and explains the ELSA and LCA services under the AgrifoodTEF service catalogue. Section 5 conveys the AgrifoodTEF service providers’ considerations on the purpose and potential implications of their services. Finally, Section 6 concludes by evaluating the findings regarding the ELSA and LCA services under the AgrifoodTEF service catalogue, outlines key insights, and provides future directions for research and practice. 2. Methodology of the Deliverable This research utilises a multi-faceted methodology to comprehensively examine the ELSA aspects in AI development within the agri-food sector. The methodology has the following two main components. 2.1 Exploring the Current State of ELSA Research (Literature Review) The first step methodology involves conducting desk research to review literature regarding the foundational understanding of ELSA. This review includes an examination of both academic and grey literature, focusing on foundational theories, frameworks, and definitions of ELSA. By analyzing key publications reached through different search engines with the search keyword of ‘Ethical, Legal, and Social Aspects (ELSA)’, this review traces the background, evolution, and current state of ELSA, with a particular emphasis on its implications for AI applications in the agri-food sector. Additionally, the review focuses on the ethical, legal, and social aspects separately to better understand the three main pillars of ELSA research - forming a basis for evaluating current practices and guiding future developments in responsible AI. We also conveyed the basics of LCA literature with a similar methodology, and discussed the interrelation of ELSA and LCA frameworks. In this regard, the deliverable also has certain features of comparative and interdisciplinary research methods. 2.2 Listing and Analyzing ELSA and LCA Services in the AgrifoodTEF Project The second part of the deliverable examines the ELSA services listed in the AgrifoodTEF project’s service catalogue, which is continuously in development by AgrifoodTEF partners that provide services. This analysis explores the purpose and expected impacts of these services by also benefiting from the insights generated through the literature review about ELSA research. This part aims to identify best practices and potential pitfalls in the implementation of ELSA services in the agri-food sector. The methodology for this analysis involved several key steps to ensure a comprehensive understanding of the services as well as their purposes and anticipated impacts. First, we listed all the ELSA and LCA services identified in the service catalogue – updated as of December 2024, at the latest. We included all services, which have ethical, legal, or social considerations or life cycle assessment in their essence. The list of services and basic descriptions are available in section 4. Next, organisations or nodes providing these services were 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 16 3. Technical Investigations: These can involve either analysis of how people use related technologies, or the design of systems to support values identified in conceptual and empirical investigations. These investigations are intended to be iterative, allowing the designer to modify the design continuously. VSD uses methods that follow closely from the theoretical constructs in the VSD literature, some VSD methods are adapted from established methods, while others are new (Fordyce 2020). For example, value scenarios built from traditional scenario-based design to better account for direct and indirect stakeholders (Benn, Abratt, and O'Leary, 2016) and systemic impacts over time. Another method, stakeholder tokens (Yoo, 2021), helps identify direct, indirect, excluded, and other stakeholders and the relationships among stakeholder groups. In practice, VSD requires a commitment to the theoretical constructs coupled with skillful use and adaptation of method in response to the complexities of the design situation. It is important to note that VSD is not just about the design of the technology itself, but also about understanding the context in which the technology will be used and the values of the people who will be directly or indirectly affected by the use of the technology. The relationship between VSD and ELSA / RRI is quite significant (Sadek et al., 2024). Both concepts aim to integrate human and ethical values into the design and innovation process. However, the characteristics of digital platforms challenge the fundamental assumptions of VSD. Digital platforms exhibit a novel form of uncertainty, namely, ontological uncertainty: even with full information and overview, it cannot be foreseen what users or developers will do with digital platforms. Hence, predictions about which values are affected might not be held. To overcome this challenge, it has been suggested (RI and VSD, 2019) to expand VSD methods to account for value dynamism resulting from ontological uncertainty. This involves extending VSD to the entire lifecycle of a platform, broadening VSD through the addition of reflexivity, i.e., second-order learning about what values to aim at, and adding specific tools of moral sandboxing and moral prototyping to enhance such reflexivity. Value Sensitive Design (VSD) can be applied in various ways in the research and design of artificial intelligence (AI) applications in agriculture. Some potential applications include (World Economic Forum 2021): 1. IoT-enabled micro-irrigation and water resources planning: VSD can be used to design AI systems that control irrigation based on the specific needs of the crops and the values of the farmers and other stakeholders (like farmers, local communities, and environmental groups), including economic but also social and environmental aspects like underground water pollution. 2. Crop-health protection: AI systems can be designed using VSD to apply pesticides, considering the health of crops, the safety of farmers, and the impact on the environment by taking into account 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 17 stakeholders' values and conflicts among them during the design and development process of the AI system. 3. AI-enabled agri-inputs advisories: VSD can guide the design of AI systems that provide advice on agricultural inputs. These systems would consider the values of the farmers, such as their economic situation, their goals, and their attitudes towards risk, in relation to other stakeholders’ values. 4. Smart Crop Insurance using Remote Sensing: VSD can be used to design AI systems that use remote sensing data to provide crop insurance and take into account not only the reduction of risks for insurers, but a more holistic view of all involved stakeholders’ values including farmers, and for instance their privacy. 5. Blockchain-enabled Fintech Solutions: VSD can guide the design of AI systems that use blockchain technology to provide financial services to farmers. These systems would consider the values of the farmers, such as their need for secure transactions and their attitudes towards technology. 6. “Uberization” of Farm Machinery: VSD can be used to design solutions that facilitate sharing of farm machinery, for instance, proposals like “Robot as a Service”. These systems should consider the values of the farmers, especially when there are barriers for them to get direct access to these technologies,, including their attitudes toward sharing, who collects and has access to the data collected during machinery operations or potential privacy issues. 7. Creation of Dynamic E-soil Health Cards: VSD can guide the design of AI systems that provide dynamic e-soil health cards. These systems would consider the values of the farmers, such as their need for accurate and timely information about soil health. In all these applications, the goal is to ensure that AI technologies not only do no harm but also contribute to good, respecting and promoting human values. It is about making AI more human-centered, inclusive, and capable of actualizing benefits across various spheres of influence. 3.5 ELSA & AI Use in Agriculture ELSA research in AI for agriculture aims to strike a balance between technological advancement and ethical responsibility. By integrating principles of responsible research and innovation (RRI), ELSA frameworks guide the development and deployment of AI technologies that align with societal expectations and contribute positively to sustainable food production. The scope of ELSA research in the agri-food sector is broad and multifaceted. It encompasses three main pillars of ELSA: i) ethical considerations such as fairness in algorithmic decision-making, ensuring AI systems do not generate biases, and promoting transparency in automated processes; ii) legal compliance considerations, especially centralising EU Artificial Intelligence Act, EU Data Act, EU and national data protection laws, intellectual property rights, and other relevant regulatory frameworks according to the specific case; and iii) social impacts evaluating the social implications of AI adoption, including effects on rural 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 18 communities, labor practices, and societal trust in technological innovations etc. Inherently, ELSA research is an interdisciplinary approach integrating insights from ethics, law, social sciences, and technology studies to provide holistic assessments of AI technologies in agriculture. The three pillars of ELSA (ethical, legal and social dimensions) are both separate and interrelated. It is a distinctive strength of ELSA that it deals at once with all three dimensions, in a way that allows researchers and providers to address often wicked problems presented by technological developments (Rittel and Webber 1973). An example of this is given in section 3.5.3, which shows the complexity of addressing the challenges labour presents within the social dimension, but also underlines that these challenges are at once legal and ethical. ELSA is, thus, an integrated approach that allows a nuanced yet comprehensive understanding of technological innovation and its effects across a wide range of contemporary realities. Still, we investigate these three main pillars of the ELSA research separately below to express in detail insights on ethical legal and social aspects. 3.5.1 Ethical Aspects of AI Technology in Agriculture In order to navigate the straights of a complex and shifting ethical landscape, ELSA follows developments in ethical debates shaped by emerging and often disruptive technologies. This is essential in order to ensure that research and innovation align with ethical norms and values undergoing significant transformation, and that potential risks and benefits are carefully assessed. The initial focus of ELSA research was on ethical concerns in biotechnology, such as genetic privacy and informed consent as explained above (Zwart and Nelis 2009). Over time, this focus expanded, adding emphasis to the legal and social aspects of issues that had been primarily considered in ethical terms, while also expanding to various unexplored techno-scientific areas such as media, defence, public safety, welfare and food system applications, among others. Moreover, ELSA research has remained close to the reconfiguration of all these areas by the introduction of AI. Accordingly, the complex ethics of decision-making processes now overlaid with AI, generate novel questions about bias and non-discrimination, algorithmic transparency and privacy, as well as technological beneficence and fairness. In the fields of agriculture and food production, the ethics of AI and digital tools bring about a plethora of issues. Many of them present difficult choices in the face of critical problems, which must be addressed by the stakeholders involved. The existing literature (e.g. Ryan 2019; Jobin et al. 2019; Ryan and Stahl 2020; Ryan 2023) highlights certain ethical issues, which may be grouped under the following indicative headings: i. Freedom and autonomy: Where tools are called to enhance the farmers’ autonomous decisions by affording novel insights and recommendations while featuring clear terms and conditions of use and allowing the modification of equipment and software and the combination with alternative technologies. 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 19 ii. Beneficence and non-maleficence: Where tools should promote trust and the general good and cause no harm to animals, plants and humans either on the farm or in the surrounding ecosystems. The focus is on whether AI and digital tools can result in harm to animal welfare and the environment. iii. Data governance: Where agricultural data ‘ownership’, privacy, access, sharing, and re-use are integral. iv. Transparency: Where tools and their outputs are expected to be interpretable and explainable and to disclose sufficient information about the uses of generated data, especially for farmers. v. Responsibility & Freedom to Switch: Where the makers of tools are expected to be accountable and assume rather than divest liability for their products. Integrity is also here important, with tools not creating path dependencies, either horizontally or vertically locking farmers into other products by the same company. vi. Digital Divide: Tools are expected to foster community and gender equity and promote a fair distribution of benefits between the producers of digital tools and the farmers. However, it is disputable if these tools are accessible to as many users as possible. vii. Sustainability: Where low energy consumption, a minimal carbon footprint and minimal use of rare earth minerals for production and use are of major importance. Also important is the tool’s easy and low-cost maintenance. Debates regarding the replacement of low-tech alternatives and the tool’s priming of biomodification are also key in this aspect. The evolution of the ELSA framework, from a narrow focus on biotechnological concerns to a broader array of scientific and technological domains, demonstrates its adaptability and ongoing relevance. Drawing on a wealth of case studies and historical insights, future ELSA research regarding AI use in agriculture will aim to enhance its anticipatory capabilities and further strengthen its interdisciplinary approach to address the ethics of emerging technologies. The aim is not merely to align research and innovation with existing ethical values and norms, but to explore the dynamic nature of such values and norms, often transformed by the very techno-scientific developments they confront. 3.5.2 Legal Aspects of AI Technology in Agriculture The legal pillar within the ELSA framework has often received less attention than its counterparts. However, it plays a pivotal role in the complex and evolving digital economy, especially considering the European Commission’s active position in regulation of the emerging digital markets in the last decade. The legal pillar of ELSA research is critical from two main points of view. Firstly, legal considerations in ELSA are essential for players in the relevant digital markets in the agri-food sector: stakeholders have to comply with the new, complicated, and plethora of legal rules. Secondly, the dynamic nature of the digital economy requires policymakers to closely monitor and take action in order to keep legal frameworks and regulations 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 20 up to date and answer the dynamic needs of the digital markets: insights generated through ELSA research and practice can be useful for policymakers to address this dynamism. We focus more on the first aspect considering the purposes of this deliverable. The evolving nature of technology presents challenges for stakeholders in understanding and adapting existing legal frameworks to address novel legal challenges in the digital age, requiring a dynamic and responsive approach to follow and comply with legal frameworks in ELSA research. Stakeholders may overlook or be unaware of critical legal aspects. Therefore, integrating compliance checks into ELSA scans acts as a safeguard, offering a systematic approach to identify and address potential legal pitfalls of the potential customers of the ELSA services. Thus, via ELSA research and practice, stakeholders can gain awareness about legal frameworks and their obligations to comply with the complicated legal framework in the digital age. Customers of ELSA services or other stakeholders are more likely to look for awareness on the intricacies and nuances embedded within these complex legal frameworks. In this regard, understanding the prominent regulations in the EU is important. The legal pillar of the ELSA research encompasses a wide range of issues that fall under various legal frameworks regarding artificial intelligence (AI), intellectual property, liability, data privacy and data governance. This section, thus, provides some insights into the different layers of the potential of the legal pillar of the ELSA research and practice. The following part of the section provides a general overview (considering the fact that deliverable D3.1 already provides an exhaustive description) of the most relevant EU regulations for the digital transformation of the agri-food sector. 3.5.2.1 Artificial Intelligence and Liability The European Commission developed a comprehensive AI package to update certain existing legal regimes and also provide new regulations (European Commission n.d.). Significant steps have been taken to regulate artificial intelligence (AI). The Artificial Intelligence Act (Regulation (EU) 2024/1689) aims to harmonise legal requirements for AI systems, focusing on safety, transparency, and accountability. It classifies AI systems into risk categories and mandates human oversight for high-risk systems. The Artificial Intelligence Liability Directive (COM/2022/496 final) adapts non-contractual civil liability rules to AI, ensuring victims receive protection and easing the burden of proof. The Machinery Regulation (Regulation (EU) 2023/1230) replaces the existing directive and addresses safety requirements for machinery, including AI-integrated components. It covers both professional and consumer products, emphasizing compatibility and risk assessment. Finally, the General Product Safety Regulation (Regulation (EU) 2023/988) replaces the existing directive, ensuring safety for all consumer products, regardless of AI involvement. The GPSR strengthens market surveillance, introduces accident reporting, and enhances recall procedures. 3.5.2.2 Data Privacy and Data Governance There are plenty of data-related regulations in the EU. The first and main data regulation in the EU is about the privacy of individuals, and in particular, about personal data protection. The General Data Protection Regulation (Regulation (EU) 2016/679) governs data protection and privacy within the EU. It establishes key 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 21 principles for lawful data processing, purpose limitation, data minimization, accuracy, and accountability. The GDPR enhances individuals' control over their data by providing them with several data rights such as the right to data access and the right to data portability, ensuring transparency and consent-based data use. It also includes provisions for cross-border data transfer and imposes significant penalties for non-compliance, influencing global data protection standards. The Free Flow of Non-Personal Data Regulation (Regulation (EU) 2018/1807) aims to remove restrictions on the movement of non-personal data within the EU, fostering a competitive digital economy. It ensures data portability and prevents data localization requirements while promoting trust in cross-border data storage and processing. This regulation complements the GDPR by facilitating seamless data use across sectors while maintaining high standards. However, there are no binding data rights in this regulation, unlike the GDPR. The Data Governance Act (Regulation (EU) 2022/868) establishes a framework for data sharing and reuse, focusing on publicly held data. It also provides obligations for data intermediaries, and promotes trust in data altruism and cross-sectoral data sharing, supporting the EU's strategy for a robust digital economy. The DGA aims to ensure neutrality in data access, enhance interoperability, and strengthen data-sharing mechanisms to benefit EU citizens and businesses. The Data Act (Regulation (EU) 2023/2407) complements the DGA by fostering a competitive data market and clarifying rights for data usage and sharing. Just like the GDPR, the Data Act is a fundamental regulation with binding data access and sharing (portability) rights for users of IoT devices. It ensures fairness in the allocation of data value, particularly for industrial and IoT data, and introduces safeguards against unfair contractual terms. By setting conditions for public sector access in emergencies and defining interoperability standards for the players in data spaces, the Data Act aims to support data-driven innovation and strengthen Europe’s digital landscape. Understanding and complying with all these regulatory frameworks are critical for the digital markets in agri-food sector because these horizontal provisions also apply sectoral data access and sharing issues (See more in Atik 2022; 2023). Potential customers of the AgrifoodTEF services are likely to seek detailed insights on the possible implications of all these data regulations in the EU. 3.5.2.3 New Legislative Framework (NLF) Adopted in 2008, the new legislative framework aims to improve the internal market for goods and strengthen the conditions for placing a wide range of products on the EU market. It is a package of measures that aim to improve market surveillance and boost the quality of conformity assessments. It also clarifies the use of CE marking and creates a toolbox of measures for use in product legislation. One of the aims of the NFL is to better protect both consumers and professionals from unsafe products being placed on the European internal market. These directives and regulations provide "essential requirements" and the detailed requirements that are defined in the so-called harmonized standards with these directives and regulations. There are Directives and Regulations under which products lying within the scope of our AgrifoodTEF project are subjected. They are often "smart devices" used in the agri-food industry and include, for example, robots automating the spraying of garden plants, systems for measuring soil composition, control and gas content controls used in animal husbandry, intelligent hives for the production of honey and others. 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 22 They use sophisticated algorithms and most often have their physical implementation in the form of electronic devices. That is why the following NFL regulations apply to the CE-marking of agrifood devices but are not limited to: • Machinery Regulation 2023/1230/EU - (instead of Machinery Directive - 2006/42/EC) • Electromagnetic Compatibility - Directive 2014/30/EU • Radio equipment - Directive 2014/53/EU • Low Voltage - Directive 2014/35/EU • Drones - Commission Delegated Regulation (EU) 2019/945 on unmanned aircraft systems and on third-country operators of unmanned aircraft systems • Restriction of Hazardous Substances in Electrical and Electronic Equipment - Directive 2011/65/EU • Pressure Equipment - Directive 2014/68/EU • Simple Pressure Vessels - Directive 2014/29/EU • Non-automatic Weighing Instruments - Directive 2014/31/EU • Measuring Instruments - Directive 2014/32/EU • Ecodesign requirements for sustainable products - Regulation (EU) 2024/1781 In conclusion, legal considerations in ELSA research are essential for addressing the ethical, legal, and social implications of emerging sciences and technologies. The AgrifoodTEF services can be very useful to help stakeholders comply with the complicated legal requirements in the EU and they may help increase awareness among stakeholders regarding legal compliance. The dynamic nature of technology also necessitates ongoing scholarly contributions and legal reforms to adapt existing frameworks to contemporary challenges. Future legal trends in ELSA research will likely focus on addressing the evolving landscape of relevant laws, and fostering international collaboration to navigate the complex legal challenges posed by emerging technologies. In brief, different layers of legal considerations in ELSA research are critical for increasing clarity for stakeholders in the agri-food sector. 3.5.3 Social Aspects of AI Technology in Agriculture The rapid advancement of technology has brought about significant social changes, influencing various aspects of human life. ELSA research plays a crucial role in understanding and addressing these societal aspects by critically examining the social dimensions of emerging technologies. ELSA research sheds light on the complex interactions between technology and society, offering insights into the potential consequences of techno-scientific advancements on individuals and communities. Moreover, ELSA research goes beyond merely understanding societal effects and actively seeks to address these effects by informing policy, shaping ethical guidelines, and fostering public dialogue. By engaging with stakeholders, including policymakers, industry professionals, and the general public, ELSA research aims to promote responsible innovation and ensure that technological developments align with societal values and norms. 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 23 The social aspects of innovation previously appraised and criticised within fields such as sociology, social theory and social anthropology, have been arguably under-thematised in past ELSA literature. RRI has been commonly seen as more attuned to social aspects, also incorporating a philosophy of science perspective on innovation (NL AI Coalition 2020). RRI has, thus, been able to address issues of “social justice, equality, solidarity and fundamental rights; a competitive social market economy; sustainable development and quality of life” (Stilgoe and Guston 2016). It is, thus, no accident that the idea of Social Labs as an inclusive research methodology was developed within an RRI context (Timmermans et al. 2020; Marschalek et al. 2022). However, as the preceding (section 3.3 and elsewhere) makes clear, ELSA and RRI converge, with ELSA Labs being aligned with the ideas developed within RRI’s Social Labs (See more at Timmermans et al. 2020). It is a distinctive strength of ELSA to place a renewed emphasis on the social along with the ethical and legal dimensions in aiming to support a theorisation of technological developments and foster responsible innovation. This does not only facilitate a comprehensive approach to existing challenges, but also offers opportunities for new angles of research and theorisation. Indeed, it is the social dimension that can make apparent certain limitations in thinking of ethical and legal aspects in isolation, expanding their scope from the microand meso to the macro-level and from the level of the distinctive artefact and its impacts to the interrelation of the human with an emerging technology as a whole (Ryan et al. 2024). It is the social dimension moreover, that shows the embeddedness and contingency of ethical values and legal principles and invites a reflection on their scope and meaning. It is worth noting moreover that economic considerations are also included within the social aspects in recent literature. Rather than seeing these considerations as externalities, ELSA re-introduces them at the heart of problems that are often wicked, that is, lacking a perfect, unambiguous solution (Rittel and Webber 1973). Solutions to such problems must often be negotiated by stakeholders with competing interests in which economic considerations are often decisive and, thus, cannot be discounted. A principal social aspect that is paradigmatic of the difficulty of perfect solutions and the complexity of the issues emerging technologies generate is labour. An overview of the problems surrounding labour in the agricultural sector will illustrate the need for thematising the social aspect as much as the ethical and legal aspects of ELSA when considering technological innovation. As agriculture has been undergoing a digital transformation, there is a call for AI and other smart tools to alleviate mundane and toilsome tasks, to make agricultural practices more appealing to a younger more diverse (gender, race, class) labour force, given that the current predominantly male population of farmers constitutes one of the oldest workforces in Europe (Directorate-General for Agriculture and Rural Development 2021; Zagata et. al. 2015; Rovný 2016), as it does in the US (Braun 2023; Belasco and Glauber 2024) and China (Liu et al. 2023). Being a farmer has become unappealing over the past decades due to reasons that certainly implicate the technological conditions of modern agriculture (mechanisation, (semi-)automation, genetic modification, use of pesticides and antibiotics, etc), which for many are not inviting, if not altogether problematic; however, 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 24 even more deterring has been a social devaluation of agriculture, with a gradual but steady shift since the second world war, seeing the perception of farmers slide from an invaluable cornerstone of food security to entrenched, reactionary agents of socio-economic entitlement and environmental harm (Langford 2022). Moreover, a web of socio-economic shifts has also meant that despite land consolidations (‘go big or go bust’) and state-of-the-art technologies, the average farmer today earns less, adjusting for inflation, than the average farmer a century earlier (Shepard 2013). In view of these pressures, it is not accidental that a significantly smaller group of people wish to enter the agricultural workforce today. Hoping that an individual tool or technology will be able to cut through this complex web of social forces and values to make agriculture appealing is at best wishful. Indeed, many, if not most, new digital technologies have prioritised efficiency gains, but this has not translated into a new profiling of agricultural labour. The issue, however, is even more complex. Another way of looking at this wicked problem is that recent efficiency gains are driven by the very premise of the current pressures of the agricultural workforce and the anticipation of the intensification of this tendency in the near future. As emigration becomes increasingly more difficult for seasonal agricultural workers and as the home population is unwilling to undertake agricultural work at the current conditions of labour, including renumeration and status, these pressures are bound to intensify (Ryan(Michael) 2023; Nguyen Thuy Anh 2023). Again, this is a social reality that technological innovation is called to respond to (Billingsley 2019; Inagaki, Lewis and Keohane 2024), but it also helps to shape, given that automation is bound to further reduce the incentives for labour emigration and for entering the agricultural workforce in general. Indeed, the opposite fear, where automation through AI leads to mass unemployment amongst agricultural communities is as prominent as the fear of labour shortages. These are only some of the concerns around the social transformation wrought by AI with regard to labour. Deskilling, as a form of loss of communal or traditional practical knowledge, or ultimately as the capacity for autonomous understanding and decision-making in the farm are arguably ethical as much as they are social concerns, insofar as they impact the way farmers are trained, their expectations from and experience of farming, the agricultural knowledge capital that a society possesses amongst its individuals (as opposed to large, multinational databanks and software), etc. One could continue layering this complex mosaic of social issues pertaining to labour, but it is already evident that a single topic seen from a single aspect already calls for dedicated research among scholars and deliberation among developers. It is equally evident that no single tool or technology can tackle such issues by itself. Similarly, no technology developer can hope to satisfy demands that are often competing, if not outright conflicting. ELSA research can, however, often through the mediation of an ELSA lab (see more in section 3.6.1 below), help developers position themselves within the landscape of social values and forces and express their vision through the products they develop. Again, the social dimension is often the subtlest, since a product must abide by the law by default, while many of its ethical implications—at least at the level of the artefact—will be easier to identify prima facie. The social dimension however entails intricate processes with multiple agents often 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 25 operating in the absence of clear normative regulatory positions. Moreover, these processes and the positions the agents occupy evolve over time. A product developed must, thus, consider not only what is socially desirable today, as opposed to yesterday, but also what will be desirable or at a minimum acceptable tomorrow. Similarly, social priorities and aspirations vary pronouncedly across cultures and nations, in a manner that cannot be overlooked by ELSA research and implementation. Both AI and agriculture are domains in which the EU wishes to constitute a distinctive paradigm next to the paradigms put forward by the US and China. Accordingly, at the intersection of AI and agriculture, a unique space of research and innovation opens within Europe. Europe is arguably the only significant global economic actor at present which is willing to resist oligopolies in both the digital technology and the agrifood sectors. It is also the only significant actor to place consistently the environment as a top priority in its agenda. This has been made explicit in the EU’s Farm to Fork Strategy as well as the EU biodiversity strategy for 2030, which aim to fulfil the European Green Deal. Accordingly, the EU has been the main driver in promoting agroecology, both in its received practice of minimal technological infrastructure, as well as in its transition to digitalisation and the employment of AI in farming. Such agendas with regard to technology and farming are related to the social pillar of the ELSA research before they are articulated in policy and law. ELSA research along with ELSA support to technology developers can be invaluable in navigating this complex and changing landscape. 3.6 ELSA Labs and Practice in the Agri-food Sector 3.6.1 ELSA Labs The Ethical, Legal, and Societal Aspects (ELSA) Lab approach represents a pioneering effort to address the multifaceted challenges posed by rapid advancements in technology, particularly artificial intelligence (AI). This approach integrates diverse perspectives and emphasizes iterative, experimental methodologies to embed ethical, legal, and societal considerations into the development and application of emerging technologies. 3.6.1.1 Emergence of ELSA Lab Approach A key feature within the ELSA Lab model is its adoption of a ’living lab’ structure, particularly in AI-focused initiatives to foster continuous interaction and co-creation among stakeholders from what is known as the Quadruple Helix framework—academia, industry, policymakers, and the public. This collaborative method aims to ensure a human-centered, trustworthy approach to AI development (NL AIC, 2022; Ryan & Blok, 2023). ELSA considerations have been instrumental in advancing research on the societal impacts of science and technology. These applications underscore the importance of public participation in addressing the societal dimensions of research, bringing together stakeholders like policymakers, non-governmental organizations (NGOs), scientists, and the general public. Stakeholder involvement is crucial, not least since much research 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 32 stakeholders can develop strategies that not only reduce environmental harm but also address societal concerns, creating a more sustainable and ethically sound agrifood system. In this section, we present a brief introduction to how LCA methodology works and how it complements ELSA in the agrifood context. 3.7.2 Life Cycle Assessment in Brief Production and consumption practices today are associated with harm to the environment. As a result, environmental policy has recently focused on transitioning to environmentally sustainable production and consumption patterns. For that, we often need to compare the environmental impacts of products and services against each other (Bouchery et al., 2024). Life Cycle Assessment (LCA) provides a quantitative approach to compiling and evaluating the inputs, outputs, and potential environmental impacts of a product system throughout its life cycle (ISO, 2006). The results of an LCA can be utilised to support decision-making by offering insights into areas such as material use, energy use and environmental emissions. The present-day LCA framework is based on a series of standards and technical reports issued by the ISO, the 14040 series (ISO, 2006). Based on these, LCA is standardised in these four stages: • Goal and scope definition – In this stage, the LCA study plan is defined with clarity and precision. No data collection or results calculation takes place during this phase. In an LCA report, the Goal and scope section enables the reader to quickly identify the precise questions addressed and the main principles applied. • Inventory analysis - This stage entails compiling and quantifying the inputs and outputs associated with a product across its life cycle. • Life cycle impact assessment - This stage focuses on understanding and evaluating the magnitude and significance of the potential environmental impacts associated with a product system throughout its entire life cycle. • Life cycle interpretation – In this stage the results from either the inventory analysis, the impact assessment, or both are evaluated against the defined goal and scope. This evaluation aims to draw conclusions and provide recommendations. In a typical LCA, the process starts with defining the goal and scope, followed by an inventory analysis. It may then proceed to an optional impact assessment before concluding with interpretation. However, LCA is a highly iterative process, which often requires revisiting earlier stages. For instance, after the initial inventory work, you may need to refine the goal and scope, revisit the inventory analysis after conducting an impact assessment, or review the interpretation at an early stage. 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 33 3.7.3 LCA in Agrifood Tech The United Nations Food and Agricultural Organization (FAO) has estimated that livestock alone contributes to 18% of global greenhouse gas emissions, with crop and livestock production responsible for 50% of the methane released into the atmosphere by human activities (FAO, 2013). Another pressing environmental issue is the disruption of the nitrogen cycle. Human extraction of nitrogen from the atmosphere, primarily for fertiliser production, surpasses all natural processes. This results in significant nitrogen emissions into surface waters. Moreover, agriculture is the economic sector with the largest demand for water and land, and it is a primary driver of land use change, such as converting forests into agricultural land (FAO, 2013). When addressing agrifood products and services, measures must genuinely contribute to practical improvements and not just shifting problems around. Thus, LCA is an appropriate tool for holistically assessing technological and managerial solutions (Notarnicola et al., 2017). A significant number of LCAs have thus been carried out in recent years to gain a better understanding of the environmental performance of the agri-food sector. LCA has been applied to agrifood systems for over thirty years and is widely considered the most appropriate method for identifying environmental hotspots with high precision (Barrett et al., 2020; Sica et al., 2024). Moreover, scholars have demonstrated that LCA is an ideal methodology for evaluating the sustainability impacts of innovative strategies and supporting decision-making in the agri-food sector. This is attributed to its interdisciplinary approach and holistic perspective on a system's interactive elements (Peña et al., 2021). The life cycle of a food product is divided into six stages: production and transportation of inputs to the farm, cultivation, processing, distribution, consumption, and waste management. Many LCA studies concentrate on the initial two stages—cradle-to-farm gate studies—because these stages often account for the majority of environmental impacts (Dijkman et al., 2018). This is primarily due to factors such as animal husbandry, manure management, fertilizer production and use, and the consumption of fuel for farm machinery operations. Figure 1 The six stages in LCA of agricultural products A challenge in the agri-food sector is the scarcity of reliable and up-to-date inventory data on food products and processes, which is essential for accurate life cycle assessment (LCA) studies and analyses, as well as for effective communication and spotting of hotspots (Notarnicola et al., 2015). This scarcity thus leads to data- 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 34 driven analyses that do not align with actual informational needs, imposing uncertainty on the results produced. This scarcity is most notable in materials such as herbicides, pesticides, and fertilisers, nitrogen and phosphate compounds, emissions and dispersion of pesticides, agricultural machinery usage, and CO2 emissions balance (Notarnicola et al., 2012). This lack of comprehensive data can result in ambiguous interpretations and conclusions. In the era of Industry 4.0, the accuracy and effectiveness of LCA analyses are increasingly crucial for strategic decision-making (Bhinge et al., 2015). In the agri-food sector, where efficiency and sustainability are critical, the integration of digitalisation, artificial intelligence (AI), and robotics can offer substantial improvements. However, these technologies also come with their own environmental footprints. LCA provides a comprehensive framework to assess the environmental impacts of these technologies, enabling stakeholders to make informed decisions about their implementation and optimisation. By systematically evaluating each stage of a technology’s life cycle, LCA can identify potential hotspots and improvement areas, thus contributing to the more sustainable integration of digitalization, AI, and robotics in agrifood systems (Notarnicola et al., 2017). However, the implementation of digital technologies involves significant energy consumption and electronic waste generation (MacPherson et al., 2022). LCA can assist in evaluating the environmental trade-offs associated with digitalization by quantifying the benefits of reduced food waste and optimized supply chains against the potential ecological costs of increased energy use and electronic waste. For instance, LCA studies have shown that digital technologies can reduce environmental impacts by enhancing logistics and resource management, leading to more efficient production and distribution processes (Belaud et al., 2019). Artificial intelligence has the potential to revolutionise agrifood systems by optimising resource use, predicting crop yields, and enhancing pest management. However, AI systems require substantial computational power, which can lead to increased carbon emissions and energy consumption. LCA can play a crucial role in balancing these environmental costs with the benefits of AI by analysing the entire life cycle of AI systems, from data centres to end-user applications. By identifying areas where energy efficiency can be improved, LCA can help developers design AI solutions that minimise environmental impacts while maximising benefits. For example, optimising algorithm efficiency and utilising renewable energy sources for data centres can significantly reduce the carbon footprint of AI applications in agriculture (Mohammadi Kashka et al., 2023). Robotics offers significant opportunities for increasing productivity and precision in agriculture, from automated harvesting to targeted pesticide application. However, the production, operation, and disposal of robotic equipment can have adverse environmental effects, such as increased resource use and electronic waste (Pradel et al., 2022). Through LCA, the agrifood sector can evaluate the sustainability of robotic technologies by assessing material use, energy consumption, and waste generation across the life cycle of robotic systems. This comprehensive assessment can guide the design and deployment of more sustainable 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 35 robotic technologies, such as using recyclable materials and energy-efficient components (Majdoubi and Masmoudi, 2021). By identifying potential areas for improvement, LCA can guide the development of technologies that not only enhance efficiency and productivity but also contribute to sustainability goals. As the agrifood sector continues to embrace technological innovations, LCA serves as a vital tool in ensuring these advancements are aligned with environmental sustainability, enabling the sector to meet growing food demands while minimizing ecological impacts (Hellweg and Milà i Canals, 2014). Embracing LCA can lead to more sustainable technological solutions, fostering an agrifood system that supports both economic growth and environmental stewardship. 3.7.4 LCA to complement ELSA The integration of LCA and ELSA offers a robust and comprehensive approach to sustainability in the agri-food sector. By combining these methodologies, stakeholders can gain deeper insights into the environmental, ethical, legal, and social dimensions of agrifood systems, enabling more informed decision-making and the development of sustainable practices. For example, an LCA might reveal that a particular farming technological solution has a low environmental impact, but ELSA considerations could highlight potential ethical issues, such as labour exploitation or negative impacts on local communities. By considering both sets of information, decision-makers can develop strategies that are not only environmentally sustainable but also socially responsible and legally sound, ensuring that solutions do not come at the expense of ethical or social values. Policymakers can also benefit from the integration of LCA and ELSA by gaining a comprehensive view of the impacts and trade-offs associated with agri-food policies. For instance, policies aimed at reducing environmental impacts through technological innovations can be assessed through LCA, while ELSA can evaluate how these innovations might affect employment, cultural practices, social dynamics, and legal frameworks. This combined approach allows for the creation of policies that are well-rounded and considerate of multiple dimensions of sustainability. Incorporating both LCA and ELSA into sustainability assessments can improve stakeholder engagement by addressing the concerns of various interest groups, including consumers, NGOs, industry players, and governments. LCA's scientific rigour appeals to those focused on environmental metrics, while ELSA's attention to ethical and social issues resonates with stakeholders concerned about human rights and social justice. By transparently communicating both environmental and socio-ethical impacts, organizations can build trust and foster collaboration with stakeholders. Integrating ELSA and LCA can spur innovation in the agri-food sector by identifying opportunities for improvement across environmental, ethical, legal, and social dimensions. For instance, innovations that 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 36 reduce environmental impacts, such as robotics and digitalisation can be further refined to enhance social benefits, like improving local livelihoods or ensuring fair labour practices. This multidimensional approach not only addresses sustainability challenges but also creates value for society and the environment. The combination of LCA and ELSA helps identify and mitigate risks associated with environmental and socioethical issues in agri-food systems. LCA can pinpoint potential environmental risks, such as resource depletion or pollution, while ELSA can highlight legal compliance issues and social risks, such as community opposition or ethical controversies. By addressing these risks proactively, organizations can enhance their resilience and ensure compliance with regulatory and societal expectations. Still, there are considerable limitations and challenges in the integration of LCA and ELSA. One of the major challenges in combining LCA and ELSA is the inherent complexity of integrating two distinct methodologies. LCA is a quantitative method focused on environmental impacts, while ELSA is often qualitative, addressing ethical, legal, and social implications. This dichotomy can lead to difficulties in aligning the two approaches in a cohesive framework. In particular, LCA relies heavily on quantitative data for accurate assessment, which is often lacking in the agri-food technology sector, particularly for emerging technologies. Data gaps can lead to uncertainties in the assessment results. On the other hand, ELSA requires qualitative data that can be subjective and variable across different cultural and legal contexts. The challenge lies in obtaining high-quality, reliable data for both LCA and ELSA components, and ensuring that they are compatible for integration. Also, there is a lack of standardized methodologies for integrating LCA and ELSA, which can result in inconsistencies in assessments. LCA has established standards such as ISO 14040, but ELSA lacks similar standardized guidelines despite individual efforts to develop internal methodological guidelines in ELSA assessments just like in the ELSA scan service (S00138) provided under the AgrifoodTEF service catalogue. This absence of standardization can lead to variations in how assessments are conducted, making comparisons difficult and potentially undermining the credibility of the results. Therefore, the integration of ELSA and LCA frameworks requires a multidisciplinary approach, bringing together experts from environmental science, ethics, law, and social sciences, which can be logistically challenging and resource-intensive. However, thanks to the AgrifoodTEF service catalogue, both services will be available to customers in the agri-food sector and potential limitations about logistics can be overcome via online communication and virtual meetings. The agri-food technology sector is subject to diverse ethical and social considerations, including food security, land use, animal welfare, and labour rights. ELSA requires the assessment of these aspects, which are often context-specific and culturally sensitive. This can lead to challenges in defining universal ethical and social criteria that can be integrated with LCA. Additionally, there may be conflicting ethical considerations, such as the trade-off between environmental sustainability and economic viability, which need careful balancing. Interpreting and communicating the results of integrated LCA and ELSA assessments can be complex. The results need to be presented in a way that is understandable and actionable for decision-makers, who may 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 37 not have expertise in both environmental and ethical-social aspects. This requires the development of clear communication strategies, shared testing protocols and metrics and the translation of complex assessment results into practical recommendations. Finally, conducting comprehensive assessments that integrate LCA and ELSA can be resource and timeintensive. This may pose challenges for organizations with limited resources, particularly small and mediumsized enterprises in the agri-food sector. Balancing the need for a thorough assessment with available resources is a key challenge. In this regard, the AgrifoodTEF services can play a precious role for the sector to serve a wide range of potential customers with different scales. 4. ELSA and LCA Services Offered within the AgrifoodTEF In this section, ELSA and LCA services are listed. We included the services in section 4.1 below by checking whether they have ethical, legal, or social considerations in their official descriptions. We also listed services, which have life cycle assessment in section 4.2 below as an extra component of the section considering the Task 3.2 title: “Collect, improve and develop ELSA and LCA services” and related descriptions of the task. The catalogue of services available under the AgrifoodTEF is regularly updated and is available online: https://www.AgrifoodTEF.eu/catalogue-of-services. As the complete catalogue of services is available online, this chapter contains only the abstracts of the services offered by AgrifoodTEF. The service identifications (Service ID) allow to find services available online, in the search bar. Each partner of AgrifoodTEF has been asked to submit a form in order to add a service into the AgrifoodTEF catalogue of services. Then, the form is duly reviewed by WP3 experts and also by WP4 (consistency with the catalogue) and WP5 (marketing and wording checks) teams. This reviewing stage is essential to ensure the harmonization and the professionalization of services. It is an ongoing work that will also apply to all newly added services. The provided list below is based on the available services as of December 2024. 4.1 List of ELSA services The services, which provide ethical, legal or social considerations to the target customers, are developed by different partners in the AgrifoodTEF project. As can be seen below, there is a strong representation of services, which have legal or standards compliance checks based on the EU regulations and/or other standards. Beyond that, there are also services, which have ethical and social aspects. Table 1 - List of ELSA Services ID Name Short description Partner Country S00007 Evaluation of conformity roadmap for an innovation. Identify necessary steps for an innovation under product development to conform regarding standards, laws and directives that apply for this specific product. RI.SE Sweden 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 38 S00009 Compliance assessment for robotic machines Help companies to understand and conform to existing regulations, machinery standards, directives and give support on how to handle situations where no clear requirements are in place, for example when using new sensor technologies as part of a safety function and for which current functional safety standards does not yet cover these aspects. RI.SE Sweden S00138 Evaluation for Ethical, Legal and Social Aspects (ELSA Scan) for responsible AI development What are the ethical, legal and social aspects (ELSA) of AI driven technology in agri-food? Mitigation of potential risks and finding opportunities are easier by identifying ELSA aspects in early development stages, when the technology is still in the making. Examples of ELSA aspects are autonomy, transparency and bias, but also data privacy and other legal aspects. How does your AI technology deal with ELSA aspects? A combined survey and interview help AI developers to identify opportunities and issues with ELSA in the agri-food context from the perspective of end users and more broadly; for society. For example, to better align with the sustainability objectives of the AI-driven technology. The outcomes of the ELSA scan are a list of identified key ELSA aspects and high-level recommendations to further improve the AI technology. WUR The Netherlands S00188 Testing for safety in alignment with relevant legislation By using state of the art test case generation empowered by Artificial Intelligence we are able to enrich testing scenarios and "think creatively" about safety in a complex agricultural system (including navigation, vision perception, etc.). The service comprises requirements revision, test cases generation, test execution and analysis of results fulfilling the need for assessment of conformity to potential regulations (within the broad AI and Robotics safety legislations). FBK Italy S00214 Conformity assessment and compliance tests Verification of a prototype in terms of evaluating the broadly understood safe use of machinery, equipment and components in the agri-food industry, the safety of conducting tests and field trials, taking into account risk analysis and other essential requirements coming from for instance New Legislative Framework directives and other EU law to better protect both consumers and professionals from unsafe products to be placed on the European internal market. One of the aims is to help manufacturers in legal placing agrifood products on EU single market and in CE-marking process. L-PIT Poland 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 39 S00216 Policy Lab The service bridges the innovation-regulation gap. We assess a company's innovation, identify regulatory hurdles, and work to brainstorm solutions and navigate regulatory bodies. This ensures their innovation thrives within a future-proof framework. L-PIT Poland S00245 User Experience (UX) Evaluation Evaluation of the User Experience of services and products in the agrifood sector UdL Spain S00310 Evaluation of ethical aspects in agrifood technology design and development We help agri-food sector SMEs developing AI and roboticsbased products and services incorporate ethical considerations into their design and development process using Value Sensitive Design methods. UdL Spain These ELSA services provide a robust and adaptable framework aimed at embedding ethical, legal, and societal dimensions into the innovation process of companies developing Robotic and AI solutions across various EU member states. With diverse expertise tailored to specific agriculture sectoral needs, these services empower technology developers to support the responsible development of their innovations to become ethically acceptable, legally compliant, and socially responsible. This broad coverage not only fosters trust and compliance but also supports the alignment of technologies with societal values. As we proceed to develop, provide and analyze these services, it is vital to assess their impact, effectiveness, and adaptability, considering the distinct contexts and requirements of different customer groups, to fully understand their potential in advancing responsible innovation within the EU. We will convey their purposes and expected impacts in detail in section 5.1 below. 4.2 List of LCA Services Two LCA services were provided from the AgrifoodTEF network, both from RISE in Sweden (Table 22). Table 2 - List of LCA Services ID Name Abstract Partner Country S00329 Life cycle assessment (LCA) of agrifood products with and without AI, Robotic & DIGitalisation (AIRDIG) services This service shows how using AI, robotics, and digital tools can reduce the environmental impact of food production compared to traditional methods, helping the stakeholders see the benefits of adopting these technologies for more environmentally sustainable operations. RISE Sweden S00330 Life cycle assessment (LCA) of AI, Robotic & DIGitalisation (AIRDIG) services in agrifood sector This service evaluates the environmental impacts of AIRDIG services throughout their lifecycle, focusing on energy use, resource consumption, and potential sustainability improvements through optimization and waste reduction. RISE Sweden 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 40 These LCA services provide a comprehensive and insightful framework for evaluating the environmental impacts of integrating AI, robotics, and digitalisation into agrifood operations. By focusing on critical aspects such as energy consumption, resource use, and sustainability improvements, these services help stakeholders understand the tangible benefits of adopting advanced technologies for more sustainable food production. Tailored specifically to the agrifood sector, these services empower technology developers and users to make data-driven decisions that align with environmental sustainability goals. A detailed analysis of these services’ purpose and their possible impact will be presented in section 5.2 below. 5. Purpose and Expected Impacts of ELSA & LCA Services under AgrifoodTEF The ELSA and LCA services within the AgrifoodTEF aim to support the responsible development and implementation of advanced technologies in the agri-food sector. ELSA services focus on identifying and addressing ethical, legal, and societal considerations to ensure that innovations align with societal values, comply with regulations, and address stakeholder concerns. LCA services, on the other hand, provide a systematic evaluation of the environmental impacts of technologies throughout their lifecycle, offering actionable insights to promote sustainability and resource efficiency. This section conveys the service providers' opinions over their services when it comes to the purposes and expected impacts of these services. 5.1 Purposes and Expected Impacts of the ELSA Services We distributed the below form to the service providers and requested them to answer the related parts of the form to identify their perceptions over their ELSA services. The below table demonstrates these insights in a structured form. Table 3 - Purposes and Expected Impacts of ELSA Services Code(s) and name(s) of the digital service(s) S00007 – Evaluation of conformity roadmap for an innovation Partner providing the service(s) RISE-Research Institutes of Sweden Name, country and brief profile of the (potential) customer(s) Manufacturers of field robots and tractors Located in Sweden or in Europe Brief description of the service(s) (including any customisations) This service sets the frame for next step services. Details of the purposes of the given service(s) S07 + S09 are to be read together. S07 is the starting point. It is our onboarding. During S07 we are exploring what kind of help the customer needs. Ex is the machine only on a field or field + road? How is the machine moved between fields? What is the intended use of the machine? Details of the expected impacts of these service(s) The goal of S07 is to pinpoint what kind of question that should be addressed or not addressed in S09 and make a roadmap. 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 41 Service execution state Not started yet Ongoing x Completed (please check the appropriate box) Code(s) and name(s) of the digital service(s) S00009 – Compliance assessment for robotic machines Partner providing the service(s) RISE-Research Institutes of Sweden Name, country and brief profile of the (potential) customer(s) Manufacturers of field robots and tractors Located in Sweden or in Europe Brief description of the service(s) (including any customisations) This service investigate how to CE-mark a machine. Details of the purposes of the given service(s) In S09 the actual work is carried out. From S07 we understand the relevant standards that applies for this machine. In S09 we are exploring each standard and investigate the demands on the machine and how these demands can be fulfilled. Details of the expected impacts of these service(s) The expected results are that the customer shall understand which standards that apply to the machine and how to do the job to be able to CE-mark the machine. Service execution state Not started yet Ongoing x Completed (please check the appropriate box) Code(s) and name(s) of the digital service(s) S00138 - Evaluation for Ethical, Legal and Social Aspects (ELSA Scan) for responsible AI development Partner providing the service(s) WR (Wageningen Research) Name, country and brief profile of the (potential) customer(s) The Netherlands, at the customer site or online if cross-border. Potential customers include stakeholders in the agri-food sector, such as farmers, agricultural technology companies, policymakers, researchers, and regulatory bodies. These stakeholders can benefit from the ELSA Scan by ensuring that AI technologies Utilized in agriculture adhere to ethical, legal, and social standards, thus promoting responsible innovation and sustainable practices. Brief description of the service(s) (including any customisations) What are the ethical, legal and social aspects (ELSA) of AI driven technology in agri-food? Mitigation of potential risks and finding opportunities are easier by identifying ELSA aspects in early development stages, when the technology is still in the making. Examples of ELSA aspects are autonomy, transparency and bias, but also data privacy and other legal aspects. How does your AI technology deal with ELSA aspects? A combined survey and interview help AI developers to identify opportunities and issues with ELSA in the agri-food context from the perspective of end users and more broadly; for society. For example, to better align with the sustainability objectives of the AI-driven technology. The outcomes of the ELSA scan are a list of identified key ELSA aspects and high-level recommendations to further improve the AI technology. Details of the purposes of the given service(s) When working on AI technology from day to day, considering the social context of stakeholders involved, but mostly the end user, may be hard to fit in. While considering ELSA aspects early on in the development cycle may provide AI developers with valuable insights to take into account without losing investment and time. For example, being interviewed about ELSA aspects may identify an issue considering legal requirements for the required data or actions to take to make sure the end user work environment is safe. Also, the AI tech 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 48 promote the adoption of environmentally sustainable practices and drive innovation in agrifood operations. The expected impacts include enhanced sustainability reporting, alignment with global climate goals, and the fostering of trust and transparency among stakeholders, thereby paving the way for a more sustainable and efficient food production industry. We provide a more detailed analysis about the purposes and expected impacts of the LCA services in the conclusion below. 6. Conclusion This document has presented the emergence, evolution and state-of-the-art of ELSA research, its connection with LCA, and provides an overview of the ELSA and LCA services offered by AgrifoodTEF, highlighting their purposes and expected impacts. These services are crucial for the ethical, legal, and socially responsible as well as environmentally conscious integration of AI and robotics in the agri-food sector. The state-of-the-art ELSA research within the context of AI solutions for the agrifood sector reflects a dynamic and evolving discipline that integrates and provides assessments for ethical, legal, and societal aspects to contribute to responsible and sustainable innovation. Rooted in frameworks such as Responsible Research and Innovation (RRI) and Value Sensitive Design (VSD), ELSA research has grown to address complex societal challenges and regulatory requirements associated with emerging technologies. Its scope encompasses ethical concerns, legal compliance, and social acceptance, creating a holistic approach to increase trust and foster accountability. The resurgence of attention to ELSA not only signifies a notable feature in research policy but also prompts reflections on the dynamic interplay between ELSA, RRI, and other competing approaches in shaping the future trajectory of research and innovation. In the agrifood sector, ELSA is closely related to Life Cycle Assessment (LCA), which provides a complementary evaluation of the environmental impacts of new technologies. In practice, ELSA Labs represent a pioneering model that includes ELSA scans and quadruple helix innovation workshops, with the aim of contributing to responsible AI development by integrating ELSA considerations into the core of AI design for sustainable food systems. The continuous (re)design process, contextual understanding, and stakeholder engagement foster an inclusive and participatory approach to addressing the complex challenges posed by AI in the digital economy. The AgrifoodTEF project has developed a comprehensive range of ELSA and LCA services (as listed and explained in detail above) to address the needs of the agri-food sector in the digital age; • The purpose of these services is to identify risks and opportunities concerning ethical, legal and societal aspects as well as environmental obligations, and guide the responsible design and deployment of AIdriven innovations in the sector. ELSA services support stakeholders in addressing issues related to ethics, legal compliance, social implications while aligning technological solutions with societal and end-user needs. The ELSA services provided under the AgrifoodTEF service catalogue underscore their critical roles in fostering responsible innovation within the agri-food sector. In particular, we observed an emphasis on 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 49 compliance with the laws, regulations and (safety and other) standards among the given ELSA services under the AgrifoodTEF while there are some services that aim to provide ethical and social considerations to the customers such as the ELSA scan service by WR, which aims to provide AI developers with identified ELSA aspects to take into account in the early stage of the technology development (S00138). Other examples of ELSA services are those provided by UdL. First, a service (S00245) which focuses on providing ethical design recommendations related to the user experience of customers. Second, another ELSA service (S00310) supports companies to take into account values along their design and development process, using approaches like Value Sensitive Design (VS) that combine techniques like stakeholder analysis, participatory design, value scenarios, or ethical reflection. LCA services focus on quantifying the environmental impacts of AI innovations, providing critical insights into the environmental impacts of food production, energy use, resource efficiency, potential sustainability improvements, and waste minimization across the technology lifecycle. • The expected impact of these services is that ELSA services are likely to empower stakeholders to identify and address ethical, legal, and societal aspects of agrifood technologies, such as sustainability, fairness, transparency, and privacy (e.g. S00310), and again particularly, compliance with regulations and standards (nearly all ELSA services, but particularly, S00188, S00214, and S00009). These services provide actionable insights through structured assessments, enabling the development of technologies that align with societal values and ethical standards. By ensuring compliance with legal frameworks or standards like CE marking (e.g. S00214) and fostering human-centric design principles through Value Sensitive Design (e.g. S00310), the services are expected to generate implications of enhancing trust, usability, and acceptance among users and stakeholders. They also help mitigate risks early in the development process, reducing costs and aligning innovations with broader sustainability and ethical objectives, ultimately contributing to responsible and inclusive technological advancements in the agri-food sector. Enhancing usability, ensuring intuitive operation, reducing errors, and improving workflow integration are other important expected impacts (e.g. S00245). The impacts of the LCA services vary. These services promote measurable reductions in environmental impacts while enhancing transparency and accountability in sustainability reporting for stakeholders (S00329). Promoting the adoption of efficient digital and robotic solutions in agriculture via helping resource optimization and waste reduction, enabling their sustainability with environmental standards, accelerating the development of sustainable agrifood, and building trust among stakeholders (S00330). The provided services are mainly focussing on ensuring the legal compliance of customers to the relevant regulations or standards. The services provided by WR and UdL also focus on the ethical and social pillars of the ELSA framework. Maybe one important point is that there is no service, which includes stakeholders workshops in its process of ELSA considerations, yet - considering that state-of-the-art ELSA research (see particularly section 3.6 above) emphasises the importance of fostering collaboration among quadruple helix stakeholders (researchers, businesses, government, citizens/consumers). Although the ELSA scan service 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 50 provided by the WR (S00138) is the necessary first step before the collaborative helix stakeholders workshops, it can be complemented by a comprehensive ELSA Lab service in future. Another highlight from the analysis of the ELSA services offered is that in some cases they have not been requested by any AgrifoodTEF customer yet. This is the case for the service about the evaluation of ethical aspects in agrifood technology design and development (S00310). From the experience of the partner offering this service, UdL, and accumulated during the roughly one year of participation in AgrifoodTEF, potential customers are still reluctant to contracting this kind of service because their focus is now on evaluating the technical aspects of their products and services. The Spanish node joined AgrifoodTEF one year later and the services offered are still recent. The focus for the next year will be on raising awareness about the ELSA services offered among Spanish companies, and to follow on the experience gathered by WR from the ELSA services they have already been able to offer to AgrifoodTEF customers. The synergy between ELSA and LCA services is pivotal for the seamless adoption of innovations like AI, robotics, and digital tools in agriculture. ELSA services provide the framework for addressing ethical issues, societal concerns and legal compliance, while LCA services quantify the environmental benefits and trade-offs of these technologies. This interrelation has a unique potential that innovations are not only environmentally sustainable, but also socially acceptable and aligned with ethical and legal standards. It is expected that these services may be complementary, because LCA services can be a deep dive or even an intervention after an ELSA Scan service (e.g. S00138) into the sustainability aspects. However, this remains to be uncovered by evaluating the services and we propose that TEF clients are involved in using and evaluating both services to address how they are complementary. Such an integrated approach may be helpful in building trust among stakeholders, thereby facilitating the adoption of transformative technologies, and ensuring advancements in AI and robotics align with societal values, regulatory frameworks, and sustainability goals. Regarding future directions, the next deliverable, D3.7 (originally D3.5): Finalized Set of ELSA Lab Services & Report About Their Performance in the Reflection Workshops, will build on the findings of this report. It will include performance evaluations based on feedback from reflection workshops and delivered and evaluated services. This deliverable will provide valuable insights into how these services are received and implemented by stakeholders. In conclusion, the ELSA and LCA services offered under AgrifoodTEF represent a pioneering approach to integrating ethical, legal, social, and environmental considerations into the development and deployment of AI and robotic technologies. By providing critical insights to customers, these services pave the way for responsible and sustainable innovation, establishing a model for the future of agrifood systems that are ethically sound, legally compliant, socially responsible, and environmentally conscious. The outcomes of the next deliverable D3.7 will further enhance the impact of these services to enhance their alignment with the related principles, while setting a benchmark for responsible innovation in the agrifood sector. 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 51 References AI Ethics Global Perspectives. (2023). Implementing the UNESCO Recommendation on the Ethics of AI. https://aiethicscourse.org/webinars/implementing-the-unesco-recommendation-on-the-ethics-of-ai AI4SFS. (n.d.). ELSA Labs. Retrieved December 18, 2024, from https://ai4sfs.org/nl/elsa-labs/elsa-labs Alvarez, M.J.R., Griessler, E., Starkbaum, J. (2022). Ethical, Legal and Social Aspects of Precision Medicine. In: Hasanzad, M. (eds) Precision Medicine in Clinical Practice. Springer, Singapore. https://doi.org/10.1007/978-981-19-5082-7_11. Anh, N. T. (2023). “Legal Analysis of EU Policies: Understanding the Binary Status of Labour Migration.” Fiat Justisia 17 (2): 163–74. https://doi.org/10.25041/fiatjustisia.v17no2.2599. Atik, C. (2022). Towards a comprehensive European agricultural data governance: Moving beyond the ‘data ownership’ debate. IIC - International Review of Intellectual Property and Competition Law, 53(5). https://doi.org/10.1007/s40319-022-01191-w Atik, C. (2023). Horizontal intervention, sectoral challenges: Evaluating the Data Act's impact on agricultural data access puzzle in the emerging digital agriculture sector. Computer Law & Security Review, 51. 105861. https://doi.org/10.1016/j.clsr.2023.105861 Barrett, C.B., Benton, T.G., Fanzo, J., Herrero, M., Nelson, R., Bageant, E., Buckler, E., Cooper, K.A., Culotta, I., Fan, S., Gandhi, R., James, S., Kahn, M., Lawson-Lartego, L., Liu, J., Marshall, Q., Mason-D’Croz, D., Mathys, A., Mathys, C., Mazariegos-Anastassiou, V., Miller, A., Misra, K., Mude, A.G., Shen, J., Sibanda, L.M., Song, C., Steiner, R., Thornton, P.K., Wood, S.A. (2020). Socio-Technical Innovation Bundles for Agri-Food Systems Transformation. Belasco E. J. and Glauber J. W. (2024). “An Aging Farm Population: Cause for Concern?” American Enterprise Institute American Enterprise Institute. https://www.aei.org/wp-content/uploads/2024/07/An-AgingFarm-Population.pdf?x85095 Belaud, J.-P., Prioux, N., Vialle, C., Sablayrolles, C. (2019). Big data for agri-food 4.0: Application to sustainability management for by-products supply chain. Comput. Ind. 111, 41–50. https://doi.org/10.1016/j.compind.2019.06.006 Benn, S., Abratt, R., & O'Leary, B. (2016). Defining and identifying stakeholders: Views from management and stakeholders. South African journal of business management, 47(2), 1-11. https://hdl.handle.net/10520/EJC190102 Bhinge, R., Srinivasan, A., Robinson, S., Dornfeld, D. (2015). Data-intensive Life Cycle Assessment (DILCA) for Deteriorating Products. Procedia CIRP, The 22nd CIRP Conference on Life Cycle Engineering 29, 396– 401. https://doi.org/10.1016/j.procir.2015.02.192 Billingsley, J. (Ed.). (2019). Robotics and automation for improving agriculture (1st ed.). Burleigh Dodds Science Publishing. https://doi-org.ezproxy.library.wur.nl/10.1201/9780429266737 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 52 Blok, V., L. Hoffmans, and E. F. M. Wubben. (2015). “Stakeholder Engagement for Responsible Innovation in the Private Sector: Critical Issues and Management Practices.” Journal on Chain and Network Science 15: 147–164. https://doi.org/10.3920/JCNS2015.x003 Bouchery, Y., Corbett, C.J., Fransoo, J.C., Tan, T. (Eds.). (2024). Sustainable Supply Chains: A Research-Based Textbook on Operations and Strategy, Springer Series in Supply Chain Management. Springer International Publishing, Cham. https://doi.org/10.1007/978-3-031-45565-0 Brandl, C., Wille, M., Nelles, J., Rasche, P., Schäfer, K., Flemisch, F., … & Mertens, A. (2019). Amicai: a method based on risk analysis to integrate responsible research and innovation into the work of research and innovation practitioners. Science and Engineering Ethics, 26(2), 667-689. https://doi.org/10.1007/s11948-019-00114-2 Braun M. (2023). “Feeding the Future: Farmers Are America’s Oldest Workforce. How Are We Preparing for the Next Generation?,” US Senate Committee on Aging, October 2023, 2, https://www.aging.senate.gov/imo/media/doc/senate_aging_farmers_report.pdf Carayannis, E. G. and Campbell, D. F. J. (2009). “Mode 3’and’Quadruple Helix”: toward a 21st century fractal innovation ecosystem. Edisciplinas.Usp.Br, 46, 201–234. https://doi.org/10.1504/IJTM.2009.023374 Chandler, J. A., Van der Loos, K. I., Boehnke, S., Beaudry, J. S., Buchman, D. Z., & Illes, J. (2022). Brain computer interfaces and communication disabilities: Ethical, legal, and social aspects of decoding speech from the brain. Frontiers in Human Neuroscience, 16. https://doi.org/10.3389/fnhum.2022.841035 De Reuver, M., van Wynsberghe, A., Janssen, M. et al. (2020). Digital platforms and responsible innovation: expanding value sensitive design to overcome ontological uncertainty. Ethics Inf Technol 22, 257–267. https://doi.org/10.1007/s10676-020-09537-z Dijkman, T.J., Basset-Mens, C., Antón, A., Núñez, M. (2018). LCA of Food and Agriculture, in: Hauschild, M.Z., Rosenbaum, R.K., Olsen, S.I. (Eds.), Life Cycle Assessment: Theory and Practice. Springer International Publishing, Cham, pp. 723–754. https://doi.org/10.1007/978-3-319-56475-3_29 Directorate-General for Agriculture and Rural Development. (2021) “Ageing of Europe’s farmers remains a major challenge in rural areas”. https://agriculture.ec.europa.eu/news/ageing-europes-farmersremains-major-challenge-rural-areas-2021-04-08_en#:~:text=3%20min%20read- ,Ageing%20of%20Europe's%20farmers%20remains%20a%20major%20challenge%20in%20rural,that% 20rural%20areas%20are%20facing. Dolan, D. D., Lee S. S.-J., and Cho, M. K. (2022). Three decades of ethical, legal, and social implications research: Looking back to chart a path forward. Cell Genomics, 2(7), 100150. https://doi.org/10.1016/j.xgen.2022.100150 European Commission. (n.d.). European approach to artificial intelligence. Shaping Europe’s digital future. Retrieved November 20, 2024, from https://digital-strategy.ec.europa.eu/en/policies/europeanapproach-artificial-intelligence ERA-NET Neuron. (2023). Ethical, Legal, and Social Aspects (ELSA) of Neuroscience. https://www.neuroneranet.eu/joint-calls/elsa/ 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 53 FAO (Ed.) (2013). Food wastage footprint: impacts on natural resources summary report. e Food and Agriculture Organization of the United Nations (FAO), Rome. https://www.fao.org/4/i3347e/i3347e.pdf Friedman, B., and Hendry, D. G. (2019). Value sensitive design: Shaping technology with moral imagination. MIT Press. https://doi.org/10.7551/mitpress/7585.001.0001 Fordyce, S. (2020). Value Sensitive Design: Shaping Technology with Moral Imagination. Design and Culture, 12, 109 - 111. https://doi.org/10.1080/17547075.2019.1684698 Forsberg, E. (2014). Institutionalising ELSA in the moment of breakdown?. Life Sciences Society and Policy, 10(1). https://doi.org/10.1186/2195-7819-10-1 Forsberg, EM. (2015). ELSA and RRI – Editorial. Life Sci Soc Policy 11, 2. https://doi.org/10.1186/s40504-0140021-8 García-Terán, J., and Skoglund, A. (2019). A Processual Approach for the Quadruple Helix Model: the Case of a Regional Project in Uppsala. Journal of the Knowledge Economy, 10(3), 1272–1296. https://doi.org/10.1007/S13132-018-0521-5 German Stem Cell Network. (2023). Ethical, Legal and Social Aspects. https://gscn.org/german-stem-cellnetwork/strategic-working-groups/elsa Hellweg, S., Milà i Canals, L. (2014). Emerging approaches, challenges and opportunities in life cycle assessment. Science 344, 1109–1113. https://doi.org/10.1126/science.1248361 Inagaki K, Lewis L, Keohane D. (2024). “Avatars, robots and AI: Japan turns to innovation to tackle labour crisis: Farmers, retailers and builders rethink business models as world’s fastest-ageing society runs out of workers.” Financial Times. Jan 22 2024: 6. https://www.ft.com/content/ad850ad2-6752-4ca7-99f64b947d0b741e ISO. (2006). ISO 14040:2006 - Environmental management — Life cycle assessment — Principles and framework [WWW Document]. ISO. URL https://www.iso.org/standard/37456.html (accessed 7.7.23) Jakobsen, S., Fløysand, A., and Overton, J. (2019). Expanding the field of responsible research and innovation (RRI) – from responsible research to responsible innovation. European Planning Studies, 27(12), 23292343. https://doi.org/10.1080/09654313.2019.1667617 Jarmai, K. (2019). Introduction. In SpringerBriefs in research and innovation governance (pp. 1–5). https://doi.org/10.1007/978-94-024-1720-3_1 Jobin A., Ienca M., and Vayena E. (2019). “The global landscape of AI ethics guidelines.” Nature Machine Intelligence 1, 389–399. http://dx.doi.org/10.1038/s42256-019-0088-2 Klimburg-Witjes, N. and Huettenrauch, F. (2021). Contextualizing security innovation: responsible research and innovation at the smart border?. Science and Engineering Ethics, 27(1). https://doi.org/10.1007/s11948-021-00292-y Langford S. (2022). Rooted: Stories of Life, Land and a Farming Revolution. London: Viking. Lauss, G., Snell, K., Bialobrzeski, A., Weigel, J., & Helén, I. (2011). Embracing complexity and uncertainty: an analysis of three orders of ELSA research on biobanks. Genomics Society and Policy, 7(1). https://doi.org/10.1186/1746-5354-7-1-47 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 54 Ligardo-Herrera, I., Gómez-Navarro, T., Inigo, E., and Blok, V. (2018). Addressing climate change in responsible research and innovation: recommendations for its operationalization. Sustainability, 10(6), 2012. https://doi.org/10.3390/su10062012 Liu J., Fang Y., Wang G., Liu B. and Wang R. (2023). The aging of farmers and its challenges for labor-intensive agriculture in China: A perspective on farmland transfer plans for farmers' retirement. Journal of Rural Studies 100. https://doi.org/10.1016/j.jrurstud.2023.103013 MacPherson, J., Voglhuber-Slavinsky, A., Olbrisch, M., Schöbel, P., Dönitz, E., Mouratiadou, I., Helming, K. (2022). Future agricultural systems and the role of digitalization for achieving sustainability goals. A review. Agron. Sustain. Dev. 42, 70. https://doi.org/10.1007/s13593-022-00792-6 Majdoubi, R., Masmoudi, L. (2021). Eco-design of a mobile agriculture robot based on classical approach and FEM creteria, in: 2021 International Conference on Computing, Communication, and Intelligent Systems (ICCCIS). Presented at the 2021 International Conference on Computing, Communication, and Intelligent Systems (ICCCIS), pp. 978–982. https://doi.org/10.1109/ICCCIS51004.2021.9397234 Marschalek I., Blok V., Bernstein M., Braun R., Cohen J., Hofer M., Seebacher L. M., Unterfrauner E., Daimler S., Nieminen M., Christensen M. V. and Thapa R. K. (2022). The social lab as a method for experimental engagement in participatory research. Journal of Responsible Innovation. https://doi.org/10.1080/23299460.2022.2119003 Martinuzzi, A., Blok, V., Brem, A., Stahl, B., & Schönherr, N. (2018). Responsible research and innovation in industry—challenges, insights and perspectives. Sustainability, 10(3), 702. https://doi.org/10.3390/su10030702 Mohammadi Kashka, F., Tahmasebi Sarvestani, Z., Pirdashti, H., Motevali, A., Nadi, M., Valipour, M. (2023). Sustainable Systems Engineering Using Life Cycle Assessment: Application of Artificial Intelligence for Predicting Agro-Environmental Footprint. Sustainability 15, 6326. https://doi.org/10.3390/su15076326 Müller, A. M., et al. (2022). "Ethical Considerations of Mobile Symptom Checker Applications." Journal of Digital Ethics. https://doi.org/10.1007/s11019-022-10114-y Myskja, B., Nydal, R., and Myhr, A. (2014). We have never been ELSI researchers – there is no need for a postELSI shift. Life Sciences Society and Policy, 10(1). https://doi.org/10.1186/s40504-014-0009-4 National Research Council et al. (1988). Report of the Committee on Mapping and Sequencing the human Genome. In National Academies Press eBooks. https://doi.org/10.17226/18430 NFDI. (2023). Section: Ethical, Legal, and Social Aspects. https://www.nfdi.de/section-elsa/?lang=en Nguyen Thuy Anh. (2023). “Legal Analysis of EU Policies: Understanding the Binary Status of Labour Migration.” Fiat Justisia 17 (2): 163–74. https://doi.org/10.25041/fiatjustisia.v17no2.2599 NL AI Coalition. (2020). Ethical, Legal and Social Aspects (ELSA) Labs Een referentiemodel voor maatschappelijke co-creatie omgevingen t.b.v. De ontwikkeling van zinvolle human centric AI toepassingen 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 55 Notarnicola, B., Hayashi, K., Curran, M.A., Huisingh, D. (2012). Progress in working towards a more sustainable agri-food industry. J. Clean. Prod., Working towards a more sustainable agri-food industry: Main findings from the Food LCA 2010 conference in Bari, Italy 28, 1–8. https://doi.org/10.1016/j.jclepro.2012.02.007 Notarnicola, B., Sala, S., Anton, A., McLaren, S.J., Saouter, E., Sonesson, U. (2017). The role of life cycle assessment in supporting sustainable agri-food systems: A review of the challenges. J. Clean. Prod., Towards eco-efficient agriculture and food systems: selected papers addressing the global challenges for food systems, including those presented at the Conference “LCA for Feeding the planet and energy for life” (6-8 October 2015, Stresa & Milan Expo, Italy) 140, 399–409. https://doi.org/10.1016/j.jclepro.2016.06.071 Notarnicola, B., Tassielli, G., Renzulli, P.A., Lo Giudice, A. (2015). Life Cycle Assessment in the agri-food sector: an overview of its key aspects, international initiatives, certification, labelling schemesand methodological issues, in: Notarnicola, B., Salomone, R., Petti, L., Renzulli, P.A., Roma, R., Cerutti, A.K. (Eds.), Life Cycle Assessment in the Agri-Food Sector: Case Studies, Methodological Issues and Best Practices. Springer International Publishing, Cham, pp. 1–56. https://doi.org/10.1007/978-3-319-119403_1 Oftedal, G. (2014). The role of philosophy of science in responsible research and innovation (RRI): the case of nanomedicine. Life Sciences Society and Policy, 10(1). https://doi.org/10.1186/s40504-014-0005-8 Penders, B., Horstman, K., and Vos, R. (2008). A ferry between cultures. Embo Reports, 9(8), 709-713. https://doi.org/10.1038/embor.2008.134 Piltch-Loeb, R. (2023). Special Issue ‘COVID-19 Vaccine Acceptance: Ethical, Legal and Social Aspects (ELSA)’. Vaccines. https://www.mdpi.com/journal/vaccines/special_issues/ethical_legal_social_COVID_19_vaccines . Popa, E. O., Blok, V., and Wesselink, R. (2020). A processual approach to friction in quadruple helix collaborations. Science and Public Policy, 47(6), 876–889. https://doi.org/10.1093/SCIPOL/SCAA054 Peña, C., Civit, B., Gallego-Schmid, A., Druckman, A., Pires, A.C.-, Weidema, B., Mieras, E., Wang, F., Fava, J., Canals, L.M. i., Cordella, M., Arbuckle, P., Valdivia, S., Fallaha, S., Motta, W. (2021). Using life cycle assessment to achieve a circular economy. Int. J. Life Cycle Assess. 26, 215–220. https://doi.org/10.1007/s11367-020-01856-z Pradel, M., de Fays, M., Seguineau, C. (2022). Comparative Life Cycle Assessment of intra-row and inter-row weeding practices using autonomous robot systems in French vineyards. Sci. Total Environ. 838, 156441. https://doi.org/10.1016/j.scitotenv.2022.156441 RI and VSD: Adapting HCI methods for responsibility and ... (n.d.). http://wp.lancs.ac.uk/hci-responsibleinnovation/files/2019/04/CHI2019_WS24_Final_Logler.pdf Rip, A. (2014). The past and future of RRI. Life Sciences, Society and Policy. 10, 17. https://doi.org/10.1186/s40504-014-0017-4 Rittel, Horst W.J., and Melvin M. Webber. (1973). Dilemmas in a General Theory of Planning. Policy Sciences 4 (2): 155–169. http://www.jstor.org/stable/4531523 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 56 Rodríguez, H., Fisher, E., and Schuurbiers, D. (2013). Integrating science and society in European Framework Programmes: Trends in project-level solicitations. Research Policy, 42(5), 1126–1137. https://doi.org/10.1016/j.respol.2013.02.006 Ryan, M. (2019). Ethics of using AI and big data in agriculture: the case of a large agriculture multinational. ORBIT Journal, Vol. 2 No. 2 https://doi.org/10.29297/orbit.v2i2.109 Ryan M., and Stahl B.C. (2020). Artificial intelligence ethics guidelines for developers and users: clarifying their content and normative implications. Journal of Information, Communication and Ethics in Society 19(1), 61-86. https://doi.org/10.1108/JICES-12-2019-0138 Ryan, M (Michael). (2023). Labour and skills shortages in the agro-food sector. OECD Food, Agriculture and Fisheries Papers, No. 189, OECD Publishing, Paris, https://doi.org/10.1787/ed758aab-en Ryan, M (Mark). (2023). The social and ethical impacts of artificial intelligence in agriculture: mapping the agricultural AI literature. AI & Society, 38, 2473-2485. https://doi.org/10.1007/s00146-021-01377-9 Ryan, M., and Blok, V. (2023). Stop re-inventing the wheel: or how ELSA and RRI can align. Journal of Responsible Innovation, 10(1). https://doi.org/10.1080/23299460.2023.2196151 Ryan, M., de Roo, N., Wang, H., Blok, V., Atik, C. (2024). AI through the looking glass: an empirical study of structural social and ethical challenges in AI. AI & Soc (2024). https://doi.org/10.1007/s00146-02402146-0 Rovný, P. (2016). The Analysis of Farm Population with Respect to Young Farmers in the European Union. Procedia - Social and Behavioral Sciences 220. 391-398. https://doi.org/10.1016/j.sbspro.2016.05.513 Shepard M. (2013). Restoration Agriculture: Real World Permaculture for Farmers. Acres U.S.A., Inc. Sadek, M., Constantinides, M., Quercia, D., & Mougenot, C. (2024). Guidelines for Integrating Value Sensitive Design in Responsible AI Toolkits. ArXiv.org. https://doi.org/10.48550/arXiv.2403.00145 Sica, D., Esposito, B., Malandrino, O., Supino, S. (2024). The role of digital technologies for the LCA empowerment towards circular economy goals: a scenario analysis for the agri-food system. Int. J. Life Cycle Assess. 29, 1486–1509. https://doi.org/10.1007/s11367-022-02104-2 Stilgoe, J., Owen, R., and Macnaghten, P. (2013). Developing a framework for responsible innovation. Research Policy, 42(9), 1568–1580. https://doi.org/10.1016/j.respol.2013.05.008 Stilgoe, J. and Guston, D. (2016). “Responsible Research and Innovation,” in The Handbook of Science and Technology Studies, fourth edition. (pp. 853-880). MIT Press: Cambridge, MA, USA. https://discoverypp.ucl.ac.uk/id/eprint/10052401/ Timmermans J., Blok V., Robert Braun R., Wesselink R. and Nielsen R. Ø. (2020). Social labs as an inclusive methodology to implement and study social change: the case of responsible research and innovation. Journal of Responsible Innovation, 7:3, 410-426, https://doi.org/10.1080/23299460.2020.1787751 Van Hilten, M., Ryan, M., Blok, V., & de Roo, N. (2024). Ethical, legal and social aspects (ELSA) for AI: An assessment tool for agri-food. Smart Agricultural Technology, 100710. https://doi.org/10.1016/j.atech.2024.100710 101100622/AgrifoodTEF D 3.6 Overview over ELSA services & description of their purpose and function for AI development in agri-food 31 December 2024 57 Van Veenstra, A. F., E. A. van Zoonen, and N. Helberger. (2021). ELSA Labs for Human Centric Innovation in AI. https://nlaic.com/wp-content/uploads/2022/02/ELSA-Labs-for-Human-Centric-Innovation-in-AI.pdf Von Schomberg, R. (2013). A Vision of Responsible Research and Innovation. Responsible Innovation: Managing the Responsible Emergence of Science and Innovation in Society, 51– 74. https://doi.org/10.1002/9781118551424.ch3 Wakunuma, K. and Jiya, T. (2019). Stakeholder engagement and responsible research & innovation in promoting sustainable development and empowerment through ICT. European Journal of Sustainable Development, 8(3), 275. https://doi.org/10.14207/ejsd.2019.v8n3p275 Wang, H., Blok, V., van Hilten, M. (2025 – Forthcoming) ELSA Labs for Responsible AI: A Novel Approach to Systematically Addressing Ethical, Legal, Societal Issues. Journal of Responsible Innovation (under review) Wilhelm, D., Hartwig, R., McLennan, S. et al. (2022). Ethische, legale und soziale Implikationen bei der Anwendung künstliche-Intelligenz-gestützter Technologien in der Chirurgie. Chirurg 93, 223–233. https://doi.org/10.1007/s00104-022-01574-2 World Economic Forum. (2021). “Artificial Intelligence for Agriculture Innovation”. https://www3.weforum.org/docs/WEF_Artificial_Intelligence_for_Agriculture_Innovation_2021.pdf Yoo, D. (2021). Stakeholder Tokens: a constructive method for value sensitive design stakeholder analysis. Ethics and Information Technology, 23(1), 63–67. https://doi.org/10.1007/s10676-018-9474-4 Zagata, L., Hádková, Š. and Mikovcová, M. (2015). Basic Outline of the Problem of the ‘Ageing Population of Farmers’ in the Czech Republic. Agris On-line Papers in Economics and Informatics. 7. 89-96. http://dx.doi.org/10.22004/ag.econ.207060 Zwart, H. and Nelis, A. (2009). What is ELSA genomics? EMBO reports, 10(6), 540-544. https://doi.org/10.1038/embor.2009.115 Zwart, H., L. Landeweerd, and A. van Rooij. (2014). Adapt or Perish? Assessing the Recent Shift in the European Research Funding Arena from ‘ELSA’ to ‘RRI’. Life Sciences, Society and Policy 10 (11) 1–19. https://doi.org/10.1186/s40504-014-0011-x