Humans acting unconsciously in a system can serve as autonomous generative agents

Generative art requires autonomous systems — but autonomy does not mean non-human. Humans acting unaware of (or uninterested in) their effect on a system generate data as validly autonomous as any algorithm. Seb Lee-Delisle’s Lunar Lander Trails demonstrated this: players of a web game navigated landers to the lunar surface, and their movement paths were recorded without their awareness. The composite image of all paths was visually beautiful — art created by many human agents none of whom intended to make art. Similarly, Aaron Koblin’s Flight Patterns visualized FAA air traffic data as flowing generative forms. The key requirement: the humans must be acting outside the system’s intended purpose, so their behavior remains self-determined.

Examples

Lee-Delisle Lunar Lander Trails: game players’ navigation paths composited into one image. Koblin Flight Patterns: FAA flight path data rendered as animated arcs across North America. Twitter metadata: word frequencies in news coverage visualized as generative flows.

Assessment

Propose a generative art piece using human behavioral data (specify the dataset, collection mechanism, and rendering approach) and explain how you would ensure the humans remain autonomous agents rather than performing for the artwork.