The technological background of the project is the emergence of new AI computer models that are able to automatically generate synthetic images that resemble real photographs (GAN models, e.g. thispersondoesnotexist.com) or that translate textual input into images with ever greater detail and sense of syntax (GPT-3 models, see e.g. openai.com/dall-e2/). This raises a number of pressing questions about how these 'artificial' image practices - through their training based on collected big (image) data - are shaped by specific existing image cultures and how synthetic images shape our visual citizenship onwards. The democratic challenge here is that everyone potentially participates in the creation of these images (including through their data), but the premises for active participation are opaque. The new AI image technologies not only add new facets to citizens' ability to produce images, but also set new limits for which images can be created (due to the bias of computer models towards particular visual expressions and subject preferences) and what meanings these images are assigned depending on how we use them and talk about them (as art, superficial entertainment, 'fake news', visual dreams and imaginations, constricting stereotypes, etc.)

The project is conducted by Centre for Aesthetics of AI Images (AIIM).