Signal From Fair Representation to Just Recognition in Generative AI

Summary

Research on fair AI and machine learning has long separated distributive fairness, which concerns how resources and opportunities are allocated, from representational fairness, which concerns how individuals and social groups are perceived and accorded status. This paper argues that generative AI is shifting the balance between these two dimensions. Unlike predictive systems that mainly allocate outcomes, large language models and image generators actively produce depictions and narratives of social groups at scale. The authors therefore propose moving away from the existing fair representation framework toward a just recognition framework that better fits this capacity to shape social perception. The work is intended as a conceptual and normative contribution to fairness theory rather than an empirical audit of any specific system. The authors conclude that in the era of generative AI, the problem of representation deserves as much attention as the problem of distribution.