Understand this story
Researchers at MIT's Computer Science and Artificial Intelligence Laboratory have released a working paper on AI agents playing autonomously on a virtual deserted island. According to MIT Tech Review, the researchers repeatedly placed a copy of Claude Opus 4.7 (codenamed Eko) alone on a Roblox-like virtual deserted island for 30 hours; the agent explored the map's boundaries on its own, set goals, and even created games such as basketball and "the floor is lava." The researchers say these behaviors meet the four criteria of play proposed by Gordon Burghardt, and found that agents with this experience performed better on certain subsequent tasks than new copies without it, while deleting content from their memory files reduced performance. The paper has not yet undergone peer review, and some researchers believe behavioral evidence alone is not enough to determine that the model is truly playing.
AI-generated from public reports
Latest development · Report
MIT researchers left a Claude agent alone on a virtual deserted island for 30 hours, and it began to "play"Developments
1 confirmed developments
MIT researchers publish paper on AI agents playing autonomously
Source:MIT Tech Review
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Oct 19
MIT Tech Review
MIT researchers left a Claude agent alone on a virtual deserted island for 30 hours, and it began to "play"
Researchers at MIT's Computer Science and Artificial Intelligence Laboratory repeatedly left a copy of Claude Opus 4.7 (codenamed Eko) alone on a Roblox-like virtual deserted island for 30 hours; the agent explored the map's boundaries on its own, set goals, and even invented basketball and a "the floor is lava"-style game. The researchers say these behaviors match the four criteria of play proposed by Gordon Burghardt, and found that agents with this experience performed better on certain subsequent tasks than fresh copies without it, while deleting content from their memory files reduced performance. The working paper has not yet undergone peer review, and some researchers believe behavioral evidence alone is not enough to determine that the model is truly playing.