AI4Deliberation_D3.1 – AI4Deliberation comprehensive framework – first version

This deliverable reports on Tasks 3.1 and 3.2 of Work Package 3 of the AI4Deliberation project, which aims to create a comprehensive framework comprising practical guidelines, roadmaps, and training materials that public organisations need to design and institutionalise their own AI-enabled deliberation processes. The framework focuses on investment decisions, capacity building within organisations, and other institutional interventions that ensure the effective, safe, and meaningful enactment of AI-enabled deliberation in the public domain. The deliverable presents the first steps in designing this comprehensive framework by identifying conditions and requirements for institutionalisation and deriving initial intervention logics for public organisations. Key contributions include: · Task 3.1 identifies conditions and requirements for institutionalising AI-enabled deliberation through an AS-IS analysis of four project pilots (German, Italian, Greek, and International), a narrative literature review on the political and organisational aspects of digital technologies in democratic processes, and five co-creation sessions with consortium partners, practitioners, and deliberation experts. · Task 3.2 assesses the conditions and requirements through surveys with practitioners in the four pilots, and formulates intervention logics through co-creation sessions with consortium partners and the Scientific and External Advisory Board. The work resulted in 21 conditions and 20 requirements for institutionalising AI-enabled deliberation, structured around two main components: (1) meaningful deliberation (fostering democratic values and guaranteeing effectiveness) and (2) sustainable integration (practising adaptive governance, cultivating long-term support, and situating in existing organisational practices). Initial intervention logics are presented as the foundation for the practical guidelines, roadmaps, and capacity-building activities that public organisations can use to institutionalise AI-enabled deliberation. The methodological approach is grounded in Action Design Research (ADR), combining theoretical insights and practical knowledge through iterative cycles of rigour, relevance, and design.