What to Do with AI Images? Towards an Epistemology of Trust

The aim of this article is to explore, within the field of visual culture, the ethical and social challenges raised by AI images, the reference imaginaries involved, and the expectations of meaning that arise. The paper will pursue the following line of argument: first, it will examine images that reflect forms of artificial intelligence according to the categories of operative image (i) and invisible image (ii), to investigate the agentive nature of AI images by asking “what do they want from us?” (Romele & Severo 2023) and relate these images to the critical literature that reflects on the concept of “image opératoire”. To elucidate the ways in which technology and the imaginary mutually nourish each other, I will draw on Castoriadis and conceive AI imaginary as both instituted and instituting (Castoriadis 1975), so as to claim a “thoughtful” nature to AI images. This removes them from the stereotypical and numbing forms in which they are usually presented and allows their instituting as well as instituted nature to be imagined in a way that can enable trust in them. This process opens the door to an ethics of the visible that is deeply rooted in the aesthetic regime and situated on an affective level, so to speak.