Envisioning Stakeholder-Action Pairs to Mitigate Negative Impacts of AI: A Participatory Approach to Inform Policy Making

Lay stakeholders are underrepresented in current AI governance: The paper finds that existing risk-based approaches to AI regulation are "commonly expert-centered" and lack input from those actually impacted by AI systems, highlighting a democratic deficit in how AI policies are currently developed.

Different stakeholders should bear different responsibilities for AI safety: The research maps mitigation strategies across various actors (companies, policymakers, civil society, etc.), demonstrating that addressing AI harms requires shared accountability rather than placing all responsibility on a single group.

Public priorities for AI risk mitigation differ from expert consensus: By surveying laypeople about which risk mitigation strategies matter most, the study reveals gaps between what experts prioritize and what affected communities consider important—insights that could reshape policy priorities if incorporated.

Policy fact sheets can make AI governance more accessible: The paper proposes presenting stakeholder insights in digestible formats designed for policymakers, suggesting that better communication tools could help bridge the gap between public concerns and policy development.

Participatory methods strengthen rather than replace expert policymaking: The authors emphasize this approach "is not targeted towards replacing policy-makers" but rather enriching policy development by centering affected communities' voices alongside technical expertise.