Continuing using Claude.ai for building algorithmic art, but focused on what comes next — training my own model on actual videos of myself painting, and building a digital half-life (or maybe a whole-life) art piece from that model. Not painting, not images…”other”. Break this piece out of the iFrame.

A comparison work is Refik Anadol’s Unsupervised at MoMA, where he trained a machine learning model on 138,000 images from MoMA’s collection and used real-time environmental data — weather outside the museum, ambient light, sound in the lobby — as noise inputs that continuously shifted the output. This (new) version uses the same structure but from a collection of one original oil painting, “7,917 Hours Ago”.