Democratising AI Fairness

The public is often excluded from AI development processes, despite bearing the brunt of their societal impact. Drawing on Habermas’ theory of communicative action, this study examines how society can meaningfully shape the fairness of AI systems. Using qualitative interviews with sixteen individuals across multiple sectors, we surface a structural disconnect between AI developers and affected communities, sustained by the absence of clear regulatory mechanisms and limited public accountability. Our key findings reveal two contrasting societal mindsets, dependent and independent, that shape how communities engage with fairness in AI. Beyond identifying barriers, we propose a co-creative approach that embeds public reasoning into AI design, development, and deployment. This model advances fairness not as a technical outcome, but as a product of inclusive, transparent, and collaborative engagement.