Generative AI and the politics of visibility

Abstract

Proponents of generative AI tools claim they will supplement, even replace, the work of cultural production. This raises questions about the politics of visibility: what kinds of stories do these tools tend to generate, and what do they generally not? Do these tools match the kind of diversity of representation that marginalized populations and non-normative communities have fought to secure in publishing and broadcast media? I tested three widely available generative AI tools with prompts designed to reveal these normative assumptions; I prompted the tools multiple times with each, to track the diversity of the outputs to the same query. I demonstrate that, as currently designed and trained, generative AI tools tend to reproduce normative identities and narratives, rarely representing less common arrangements and perspectives. When they do generate variety, it is often narrow, maintaining deeper normative assumptions in what remains absent.