Archaeologists from Leiden University have used artificial intelligence for the first time to reconstruct the rules of an ancient board game. A neural network analyzed geometric patterns and wear marks on a stone board found during excavations at a Roman settlement in what is now the Netherlands and determined how the pieces moved across it.

The mysterious artifact was discovered with lines etched into it and characteristic abrasions reminiscent of the wear and tear caused by the movement of gaming pieces. Researchers hypothesized that the stone was part of a board game set, although such finds had not previously been recorded in the region.

To test their hypothesis, the scientists used machine learning. The algorithm compared patterns on the stone with the rules of little-known and forgotten ancient games, using uneven surface wear as a clue to understanding the logic of the moves.

The results were unexpected: the AI identified a type of blocking game—a game in which victory is achieved by limiting the opponent's moves. Previously, historians dated the emergence of such strategy games to the Middle Ages, making this discovery the oldest known example of this genre.

"This is the first time that AI-powered game simulation has been used in conjunction with archaeological methods to identify a board game," noted archaeologist Walter Christ. He added that the new method will make it possible to reconstruct the entertainment of ancient cultures even in the absence of written sources or images.

The discovery corrects our understanding of the leisure activities of the Roman Empire and demonstrates that complex strategy games appeared in Europe much earlier than previously thought.