AI Racial Bias

Download PDF
Open Webpage

Anna M. Górska, Dariusz Jemielniak

Published: 30 Oct 2024, Last Modified: 08 Jan 2026CrossrefEveryoneRevisionsCC BY-SA 4.0
Abstract: This chapter studies the interplay between technical configurations and societal constructs within the domain of artificial intelligence (AI), with a specific emphasis on AI-enabled text-to-image generators. Drawing on conceptual frameworks informed by critical race theory and complemented by insights from science and technology studies, the discourse critically engages with the question of whether these generators inadvertently perpetuate or challenge prevailing racial stereotypes and biases. The analysis unravels the visual portrayals of four professionals in the disciplines of law, medicine, engineering, and scientific research through the lens of the nine most popular AI text-to-image generators, revealing the degree to which AI-based tools reflect and amplify societal racial biases. To evaluate in terms of race the AI-generated images of professionals, we surveyed 84 participants. The empirical evidence showcased the inclination within all AI systems towards the portrayal of white individuals, a trend that persisted even amidst a diversity of generated images. The findings show the racial bias of AI tools and its preference for “whiteness.”