<aside> âšī¸ This presentation at PAIRS 2026 Online on 17th February 2026 12:15 UTC. Registered participants will receive zoom links to join the session via e-mail.
đī¸ Thread on PAIRS Discussion Server (Discord) (register first)
</aside>
Over the past few years, participatory processes have started being employed by frontier AI companies to democratize AI development and governance processes through the inclusion of public voices. These initiatives are commonly framed in utopian terms by academics and participatory AI researchers: as "good place" visions of desirable futures in which technologies are developed and governed inclusively. However, a core tension exists between these utopian visions and current political-economic realities. In practice, the pipeline for participatory AI is broken. While utopian visions are abundant and one-off pilots are increasingly common, true institutionalisation and standardisation remain rare or non-existent. As a result, most of these visions often function as "no-place" utopias: compelling ideals that remain detached from the institutional conditions required to realize them.
By mapping participatory processes across the AI development pipeline and situating existing participatory AI initiatives by Anthropic, OpenAI, and Meta within this pipeline, we highlight that most participation currently occurs downstream during post-training or testing, while the most consequential upstream decisions regarding what to build and why remain under unilateral corporate control. Recognising this reality, we conclude by proposing a choice for the field: either fix the pipeline by finding more effective ways to mandate and institutionalise binding participatory power, or recalibrate our demands from frontier AI companies to focus on targeted, purpose-specific participatory processes.