The AI gaze and imagined places: Co-creation and negotiation with the generative AI assemblage

Generative artificial intelligence (GAI) is increasingly used to produce images of places, yet how users negotiate representations of real locations remains understudied. Building on the tourist gaze, we propose the AI gaze: a technologically mediated way of seeing place that is co-produced through negotiation between users and the socio-technical assemblage underlying GAI. Combining semantic image analysis, interviews, and 523 prompts from 36 participants in Poznań and Birmingham, we show that these images reinforce stereotypical tropes, lack diversity, and include visual glitches. We argue that the AI gaze is distinct because of how agency is distributed: the GAI-assemblage seemingly resists attempts to align images with human understandings of place, while also enabling users to remix and challenge stereotypes.