his deliverable reports on Task 2.1 of Work Package 2 (Design AI-enabled Deliberation Methods) of the AI4Deliberation project. It presents a comprehensive overview of the legal and ethical issues related to the project's implementation, covering compliance with research participant conduct, personal data processing (including special categories), and the use of AI tools deployed in deliberation platforms to support democratic processes. The deliverable includes a detailed and updated Data Protection Impact Assessment (DPIA) under Article 35 of the GDPR, and an Ethics Impact Assessment of eight groups of AI tools likely to be used in the project: Large Language Models, Video Large Language Models, Deliberation and Argument Mining Tools, Moderation and Safety Tools, Translation and Accessibility Tools, Gamification and Engagement Tools, Summarisation Tools, and Clustering Tools. These are assessed against the High-Level Expert Group on AI (HLEG) Ethics Guidelines for Trustworthy AI. The analysis covers surveys, observational research, and four pilot studies conducted in Germany, Italy, Greece, and internationally (UK-led), addressing topics including public consultation on legislative proposals, climate adaptation policy, and Long Covid.
Završník et al. (Tue,) studied this question.