The question
What happens to authorship when a machine trained on Caravaggio, Kapoor and Koons generates an image that recombines all three? Is this creativity or retrieval?
Research essay · AI & visual culture · 2023
Art, Theory & AI
A dialogue on art theory and machine intelligence, built from Midjourney prompts.
2023Year
Research essayType
AI · visual cultureContext
MidjourneyTool
A dialogue on the collision of artistic tradition and machine intelligence: what it means when AI systems trained on the whole history of human visual culture begin generating images that no single human could have made.
Prompt
Anish Kapoor’s bean meets the Titans.
Three questions
An archive that thinks back.
Visual culture
AI image generation is not neutral. It inherits the biases, aesthetics and power structures of its training data. The archive is political.
The practice
AI tools as part of an ongoing research practice: testing where they expand creative capacity and where they flatten cultural specificity.
Chapter 1
Sculpture at the scale of a city.
The same sculptural idea, pushed through prompts: arches and portals, giant beans among crowds and shops, the sculpture asking to be walked through.
Chapter 2
When the sculpture wakes up.
Push the prompt further and the archive starts to speak back: eyes, mouths, faces, tantrums. Tools trained on everything we made begin to mirror our fears.
Chapter 3
Then, a landscape.
Scale the idea up to parks, oceans and other worlds: hundreds of beans in a field, whales in glass, jellyfish among cosmic pillars.
Is this creativity or retrieval?