A study published in 2025 in Antiquity reports that researchers used artificial intelligence to reconstruct the rules of Ludus Coriovalli, an ancient Roman board game dating to roughly 1,800 years ago, based on a limestone fragment found in Heerlen, the Netherlands.

The game, described as asymmetric, pits four hound pieces against two hare pieces in a blocking strategy, and the geometric pattern incised on the stone suggested a playable board.

Microscopic analysis of wear marks on the limestone identified repeated contact points consistent with movement of game pieces, and AI simulations matched these patterns to a blocking game in which four pieces cooperate to trap two opponents. The researchers concluded that the stone represents a functional board game rather than a decorative object.

The study highlights the potential of artificial intelligence to expand understanding of ancient leisure activities, while the precise cultural role of Ludus Coriovalli remains to be explored.