Here, too, generative AI is proving to be astonishingly connected to the history of artistic practice as here, too, essential moments lie in the unknown: we do not know in detail and we can never know conclusively what happens in the many layers of artificial neural networks. The moment a diffusion model starts detecting an image in the noise of latent space – i.e. in all higher-dimensional possibilities – can be compared with the moment in which a painter stands in front of an empty canvas and projects their imagination onto the empty surface to give it form with paint and brush.
Minimal art artist Robert Ryman, who only painted white pictures, called the white surface “total sensitivity”. In his spirit, the latent space of AI could be described as the “total sensitivity” of the machine. Training an AI with images from the past would, correspond to an artist’s art-historical education, experience and recollected impressions.
3. Écriture automatique as an artistic strategy
A century ago, the surrealists put an artistic strategy called écriture automatique, spirit writing or psychography into practice. The idea was to create text and image from the unconscious, without intellectual control. This approach has a modern equivalent in the use of generative AI for art production. (The fact that one of the best-known text-to-image programs, DALL-E, is named not only after the lonely robot WALL-E but also after the most famous surrealist painter Salvador Dalí would appear to be no coincidence). Dalí saw himself as a medium, shifting the authorship of his art to his unconscious and his dreams. This did not diminish his influence as an artist. Ever since, avant-garde has no longer been the craftsmanship and personal expression of an artist but rather the idea and inspiration that is seen as the essence of an artwork.
