Going public: the role of public participation approaches in commercial AI labs

Participatory AI remains largely superficial in commercial labs: While industry practitioners view public participation as important for "societally beneficial" AI systems, the current implementation gives participants no actual decision-making power and has "little ongoing impact for AI labs," suggesting participation is more symbolic than substantive.

Cost and resource constraints are major barriers: Commercial AI labs cite participation as "expensive and resource intensive," indicating that financial limitations—rather than philosophical disagreement—are preventing meaningful public involvement in AI development at the companies driving major advances.

Participation efforts are fragmented and uncoordinated: The research found that participatory approaches are "atomised" within companies, meaning different teams operate independently without standardized methods, resulting in inconsistent and ineffective engagement practices across organizations.

Lack of transparency creates perverse incentives: There is "no incentive to be transparent about" participation adoption, suggesting companies may underreport or downplay their engagement efforts, making it difficult for the public or regulators to assess whether genuine participation is actually occurring.

The field lacks clear guidance on implementation: The absence of standardized methods and "unclear context" for how participation should work in practice leaves commercial labs without a roadmap, contributing to the piecemeal approaches that currently characterize the field.