Generative AI: a problematic illustration of the intersections of race, gender and class

This commentary investigates how generative AI tools such as DALL-E can create imagery which (re)produce racist, gendered and classist representations of peoples. Drawing on prompts entered across three time periods into DALL-E, I employ algorithmic coloniality as a conceptual framework, together with critical visual analysis, critical race semiotics and intersectionality to examine the images created. This analysis shows that, despite advances in photorealism, DALL-E persistently reproduces the same racialised and gendered tropes in its depictions of Black American women. I argue that these images are not merely aesthetic by-products, but socio-technical artefacts shaped by historically racist training data. Moreover, I suggest that enhanced photorealism may amplify, rather than mitigate, such stereotypes. Building on these findings, I argue that geography educators need to cultivate forms of critical AI literacy that extend beyond refining prompts to interrogate the algorithmic coloniality embedded within these systems. I conclude by proposing practical and collective strategies to support geography educators engaging with these tools.