Beyond Participatory AI

Abstract

The ‘participatory turn’ in AI design has received much attention in the literature. In this paper, we provide various arguments and proposals to move the discussion of participatory AI beyond its current state and towards stakeholder empowerment. The participatory AI literature points to Arnstein’s understanding of ‘citizen power’ as the right approach to participation. Although we agree with this general idea, we argue that there is a lack of depth in analyzing the legal, economic, and political arrangements required for a genuine redistribution of power to prioritize AI stakeholders. We highlight two domains on which the current discourse on participatory AI needs to expand. These are (1) the legal-institutional background that could provide ‘participation teeth’ for stakeholder empowerment and (2) the political economy of AI production that fosters such power asymmetries between AI developers and other stakeholders. We conclude by offering ways forward to explore alternative legal arrangements and ownership models for participatory AI.