Why CUA brushstroke-by-brushstroke AI art is harder to dismiss than diffusion slop

akbirthko · x · 2026-09-06

In a thread criticizing one-shot Astra demos (feedback unit, feedback richness, feedback loop), akbirthko argues that AI art made by a computer-use agent laying down brush strokes is harder to dismiss as "not art" compared to diffusion slop — even if the model did all the work — because the stroke-laying process exists.

Original post →

More from AGI Musings

AGI Musings channel →