The Maivia Gazette

Verified AI news, every morning

Research

Stratego AI Ataraxos beats the game's most decorated player 15 wins to one

Researchers from Carnegie Mellon, NYU, Stanford and MIT report in Nature that training reportedly cost less than $8,000.

A low-angle close-up of red and blue face-down game pieces on a wooden board under a single spotlight.
AI-generated illustration, not event photography. The motion is AI-generated from the still.

Researchers from Carnegie Mellon University, New York University, Stanford University and MIT have introduced Ataraxos, a Stratego-playing AI described in a paper published in Nature. The authors say that, to their knowledge, it is the first AI to reach superhuman performance in the board game. In an official 20-game series, Ataraxos beat Dutch player Pim Niemeijer with 15 wins, one loss and four draws. Training reportedly cost less than $8,000. Niemeijer is considered the most decorated player in Stratego's history. He has won four world championships, 15 Dutch national titles and two online world championships, and spent more than 600 weeks at the top of the world rankings. Stratego is a hard test for AI because each side sets up its 40 pieces face down, and the paper puts the number of possible setups at more than 10^33. In imperfect-information games like this, the value of a move depends not only on how play continues but also on what the opponent believes and conceals. That makes the game a different challenge from perfect-information games such as chess or Go. The result closes one of the last major board games where humans still led, and the reported training cost suggests strong play under hidden information no longer requires large compute budgets.

Sources

  1. The DecoderAI beats Stratego's greatest player, ending one of the last human strongholds in board gamesPublished · fetched

Also in this edition