With the widespread popularity of AI image generation tools, we have entered a creative landscape that is increasingly automated. In this new era, it is important to recognize that images that are created through AI tools can only reflect back the visual content they are trained on. In a way, the images we receive back from AI generation tools serve as aggregates of all it ‘knows’ of the visual world. Importantly, these mirrors of our visual world can include and reinforce tropes of visual style and subjects, as well as latent biases present in the art historical canon. Instead of passively consuming these AI reflections, we may instead approach them with a critical eye to better understand how visual meanings of art and media are produced, circulated and codified in AI systems. This visual essay illustrates and critiques how AI tools ‘learn’ and reproduce collective understandings about visual art and media worlds by presenting an exhibition of artificially generated ‘artworks’ derived from a set of fictionalized artist statements that represent common movements, issues, media and practices of historical and contemporary visual art. In this visual essay, the images produced act not only as artworks but also as sites of inquiry. By critically investigating and analysing how the AI generator constructs each image to represent these fictional artists, new insights may be revealed on how these technologies visualize the visual art world itself, and our knowledge of the art world may be advanced.
Brad Olson (Wed,) studied this question.