In plain words: A coding assistant was given only a virtual body on an unknown island, with no task or reward, and left to act. Across thirteen agents it spontaneously climbed, built, drew, and ran physics tests, learning new tricks — behavior that fits classical ideas of play.
Abstract · Is this machine playing?
We placed a modern AI coding assistant in an unintended role: as the mind of a body on an unknown digital island. With only a minimal instruction mentioning no specific task, reward, or activity, the machine started animating its virtual body. Across thirty-hour runs, the embodied AI agent climbed hills, stacked blocks into towers, drew mandalas, reinterpreted sports, ran experiments on the physics of its world, and learned techniques that later expanded what it could accomplish. These activities recurred across thirteen agents but diverged into distinct histories. We examine whether this behavior satisfies classical criteria for play and ask whether play can become a mode of machine development.
Nathan Cloos, Antonio Norelli, Daniel Durbin, Jacob Andreas, Daniela Rus, Phillip Isola
arXiv:2610.07130 · cs.AI, cs.LG · submitted Oct 5, 2026
abstract · pdf · html · 13 pages of main text, 17 figures