A first in a hidden-information game
Stratego poses a challenge that chess does not: the value of an opponent's pieces is hidden from players. That has made it a long-standing test for AI. In a new study, a system managed to defeat the best Stratego player in history in this imperfect-information setting.
Running on a budget
The researchers say the system achieved this result not only with powerful hardware but also with a relatively modest compute budget. That suggests the approach can work with limited resources, not just in giant data centers.
Why it matters
Decision-making under incomplete information defines many real-world problems: negotiation, security, logistics and strategy. Games like Stratego are used to test how AI can handle such uncertainty. The study was published in Nature, with technical details available in the arXiv paper.
