Scientists might have discovered what makes AI so effective at chess
AI’s comfort with complexity could be key to its chess-playing prowess
In 1997, artificial intelligence defeated the world's best human chess player, marking the beginning of AI's strategic superiority. Deep Blue, a computer system, used a brute-force approach to evaluate numerous moves. However, researchers at Complexity Science Hub in Vienna are searching for what makes AI so effective at chess. They suggest that AI's comfort with complexity could be the key to its success.
The team analyzed games from online databases, measuring "strategic tension"—the number of mutual interactions between opposing pieces. Up to the 20th move, human games showed higher average tension than AI games. After that point, human tension drops below AI's, which remains high. As the terrain becomes unfamiliar, people tend to simplify their piece arrangements due to cognitive limitations.
Higher-rated human players maintained greater tension throughout their games, while tension increased with deeper search trees used by computers. While higher tension correlated with better ability in both humans and AI, the study did not prove causation. Servedio warns that high tension in AI systems can lead to instability, making them prone to errors.
Chess remains a valuable model for studying cognition in humans and machines, with potential applications extending to high-stakes situations like diplomacy.
Written by urgent.news from Scientific American's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
