EVE OF DIFFUSION (Mona Lisa is drinking wine with da Vinci) by Merzmensch is a series of text-to-image AI artworks that explores the cultural knowledge and creative capacity of DALL-E 1. Human dreams are made up of emotions, experiences, feelings, and ideas—but what about machine dreams?
Since 2016, Merzmensch has been exploring the creative side of generative AI by experimenting with the most accessible models and approaches. Since 2020, he has been a creative member of the OpenAI Community Ambassador group. Made in 2021, on the eve of the diffusion models, he gained exclusive access to DALL-E 1 and began his usual approach: to unleash the creativity of AI by pushing models to their limits.
"Working with DALL-E 1 (based on the GPT-3 model), I discovered a deep AI awareness of human culture, contexts, and knowledge. What happens when we put these ideas together? What happens when LA GIOCONDA toasts her creator with a glass of wine?
We experience a series of unique moments—from boring overfitting (de-auratized, in Walter Benjamin's view) to the incredible staging of Lillian Schwartz's hypotheses about the Mona Lisa being a self-portrait by Da Vinci. For me, this was the beginning of a long, never-ending journey into meaningful dialogue with visual AI—after five years of inspiring but desemanticized wanderings in the semi-lit labyrinths of latent spaces of Deep Dream and GANs."
– Merzmensch