Decolonial Prompting: Rewriting AI Toward Black Futures
Abstract
This presentation develops a theory of decolonial prompting as a method of engaging large language models (LLMs) and other AI systems from the perspective of Black studies and decolonial thought. It argues that prompting is not a neutral technical skill but a political act situated within the coloniality of power, where race, knowledge, and humanity have been historically organized through Eurocentric hierarchies. Drawing on Quijano, Mignolo, and Wynter, the talk traces how modern AI systems inherit and reproduce colonial logics in their training data, defaults, and interfaces. It then introduces decolonial prompting as a practical method for Black scholars and communities to contest erasure, expose algorithmic anti-Blackness, and rewrite machinic outputs toward Black futures. Using case studies from my experience with OpenAI’s Sora and Canva, alongside mainstream chat and image models, the presentation shows how visual and textual outputs often encode Blackness as deficit or pathology. It maps the layered subject positions involved in prompting such as colonized subjects, programmers, validators, and corporate/state actors—and argues that every prompt is a negotiation between subject, object, and frame. The slide deck was prepared for the African Studies conference panel on AI, race, and decoloniality with the full paper published by the Cambridge University Press in the African Studies Review in 2026. It is intended both as a conceptual intervention and a practical resource for those seeking to engage AI critically and strategically, and in solidarity with Black and other marginalized communities.