The Orthogonal Bet Joel Simon on "Beyond Slop"
Mar 18, 2026
Joel Simon, creator of Artbreeder and researcher mixing art with computational experiments, discusses evolutionary algorithms and why smooth latent spaces matter. He recounts Artbreeder’s origins, making generative tools playful and communal. He also outlines his essay Beyond Slop and argues for tools that reveal craft and expand self-expression.
AI Snips
Chapters
Transcript
Episode notes
Unexpected Power Of Mixing Many Components
- Artbreeder enabled mixing hundreds of semantic components, producing novel extrapolations beyond single-class GAN outputs.
- Users discovered powerful tricks like negative values in the UI that produced unexpectedly strong images.
Treating A Tool As A Collective Artwork
- Joel framed Artbreeder as a collective artwork and made outputs public domain to encourage remix and downstream creative use.
- Early tensions over ownership surfaced as users wanted to sell or claim images, intensified by the NFT wave.
Why Evolutionary Algorithms Captivate Creators
- Joel's interest in evolutionary approaches came from sculpture, topology optimization, and biology showing emergent, non-intuitive forms.
- He valued systems that surprise creators and produce designs humans wouldn't conceive unaided.