Fruit fly beats advanced AI at chess in just 11 moves — and that’s just the start

A digitized fruit fly brain, mapped by HHMI Janelia Research and Google Research into 166,000 neurons and 25.6 million connections and released on Sept 3, beat Anthropic's Claude Opus 5 at chess in 11 moves in a project by creator Maxime Labonne. The connectome data has since powered dozens of viral experiments, including Stonkfly, a $100 crypto-trading portfolio with simulated dopamine reinforcement, a fighting game in which a Stanford student lost all 12 bouts to the fly-controlled opponent, and a blackjack program called FlyJack. Other users wired the neural network into animatronic cats, crab-like walkers, and small drones.

Fruit fly beats advanced AI at chess in just 11 moves — and that’s just the start See more of our coverage in your search results. Add The New York Post on Google https://www.google.com/preferences/source?q=nypost.com AI may not be so superintelligent after all. The lowly fruit fly just beat one of the world’s most advanced AI models at chess. That’s just one of dozens of wild experiments that have gone viral since scientists fully mapped the neural pathways of an adult male fruit fly — whose entire brain fits on the head of a pin — and posted the data online. The threshold-crossing technology, achieved by HHMI Janelia Research in Virginia and Google Research, catalogued all 166,000 individual neurons and 25.6 million connections in the fly’s brain and made it downloadable on Sept 3. Inventive nerds then used the digitized neural network to power and navigate a wide variety of contraptions, apps and games. Creator Maxime Labonne hooked the mapped pathways up to a chess engine https://huggingface.co/spaces/mlabonne/chessfly and pitted it against Claude Opus 5, which is one of the most advanced AI models on the market. The fly delivered a checkmate in just 11 moves. One engineer created a project called Stonkfly https://github.com/nftechie/stonkfly — giving the entity a $100 portfolio and creating a way for it to trade crypto currency. The program simulated dopamine rushes for the fly when it made money on a trade for positive reinforcement. A Stanford student created a fighting game https://x.com/poetengineer /status/2099326466714476750?s=20 where she controlled one digital fly and the neural network controlled another digital fruit fly — and battled it in a king-of-the-hill style competition. She lost all 12 bouts. The fly has also hit the casino in a program called FlyJack. One engineer created a blackjack game for the fly-brain bets against a dealer. But the fly’s success varies — like all gamblers — each time you run the experiment. Other users plugged the neural network into a variety of homemade robots, such as animatronic cats, crab-like walkers, and small drones.