News

Board games have always been a playground for AI researchers. With clear rules, defined outcomes, and structured gameplay, they're the perfect testing ground for AI systems. But what happens when game rules have glitches? Enter Diego Perez Liebana and Raluca Gaina, computer scientists at Queen Mary School of Electronic Engineering and Computer Science, who are using AI to help game designers perfect their creations.

Their startup, Tabletop R&D, uses a resource-efficient AI approach called Monte Carlo tree search to simulate thousands of gameplay scenarios in seconds. This allows designers like Alan Wallat (creator of Sirius Smugglers) to identify and fix issues, like infinite gameplay loops, before launching their games.

The AI doesn't just play to win: it's trained to mimic human behavior, from beginner instincts to strategic mastery. This makes it an invaluable tool for refining game mechanics, ensuring fairness, and eliminating dull gameplay.

See the full newsitem in the Economist: www.economist.com/science-and-technology/2025/02/26/how-a...