Stakeholder Involvement for Responsible AI Development: A Process Framework

Stakeholder involvement (ShI) is increasingly promoted for responsible AI development; however, we lack insights into the actual process that practitioners have to complete when conducting such ShI supporting responsible AI efforts. Bridging the gap between theory and practice, this work presents a process framework of ShI for responsible AI development, formalising its stages and associated challenges. We derived an initial framework by relating literature from ShI in healthcare to responsible AI development and expanding it through AI practitioner insights obtained through semi-structured interviews (n=10). The resulting process framework enables systematic reflections about ShI for responsible AI in practice: the required stages, their order and nature, associated bottlenecks, as well as promising interventions. This is essential for informing future research and further supports practitioners by facilitating more systematic communication and ShI efforts. We recommend applications of the framework to advance (our understanding of) ShI for responsible AI development in practice.