With most information hidden, the game Stratego had stumped AI—until now

A team of researchers from Carnegie Mellon, MIT, NYU, and Stanford has developed an AI called Ataraxos that has successfully defeated Pim Niemeijer, the top Stratego player, winning 15 out of 20 games. Stratego, a game of hidden information and bluffing, had previously resisted AI mastery due to its complexity and the vast number of possible piece arrangements. Unlike other games like chess or poker, Stratego involves 40 pieces with hidden identities, making it a challenge for AI. Ataraxos overcame this by using a second neural network to predict the identity of hidden pieces, a breakthrough that allowed it to excel where others, including DeepMind’s DeepNash, had failed. This achievement highlights the potential for AI to tackle complex problems with hidden variables. QUESTION: How might the development of AI like Ataraxos influence the way we approach problem-solving in real-world scenarios with incomplete information?