AI's Chess Superiority Linked to Sustained Strategic Complexity

Researchers have found that artificial intelligence maintains higher strategic tension in chess games compared to humans, especially after the first 20 moves. While human players simplify their positions as complexity increases, AI systems like Stockfish and Leela sustain intricate piece interactions throughout the game. This ability to handle complexity may explain why AI outperforms humans in strategic depth and long-term planning.
In November 2026, a study published in APS Open Science revealed that artificial intelligence systems like Stockfish and Leela maintain significantly higher strategic tension in chess games compared to human players after the first 20 moves. Strategic tension—defined as the number of potential mutual interactions between opposing pieces—peaked earlier in human games and then declined, while AI games sustained high levels throughout. This divergence suggests that humans simplify positions under cognitive load, whereas AI systems, unburdened by such limitations, preserve complex dynamics. The research, led by physicist Vito D. P. Servedio at the Complexity Science Hub in Vienna, analyzed thousands of human vs. human and AI vs. AI games from online databases, measuring tension at each move. Higher-rated human players and those playing classical time controls maintained greater tension, indicating that time and skill correlate with complexity retention. The study also found that deeper tree searches by Stockfish increased tension, though it did not establish causality between tension and superior performance. These findings help explain why AI outperforms humans in long-term planning and strategic depth, despite using similar underlying mechanisms like neural networks and tree search. The implications extend beyond chess: Servedio warns that high-strategy systems may be unstable, risking catastrophic outcomes if errors occur—raising concerns for applications in diplomacy or conflict resolution.
The study’s insights into AI’s ability to sustain complexity could influence how autonomous systems are designed for high-stakes domains like international negotiation or military strategy. While humans simplify under pressure, AI does not, which may lead to more optimal decisions—but also greater risk of instability if conditions shift unexpectedly. This mirrors earlier concerns around algorithmic trading in finance, where automated systems optimized for speed and complexity sometimes triggered market crashes. The chess model offers a controlled environment to study such dynamics, potentially informing safeguards in real-world applications. However, it remains unclear whether high tension directly causes superior outcomes or merely reflects deeper analysis. As AI systems grow more embedded in strategic decision-making, the balance between complexity and stability will become critical. What’s next may involve developing hybrid models that combine human intuition with AI’s endurance of complexity—potentially reducing risk while preserving advantage. The broader takeaway is not that AI is inherently better, but that its different cognitive profile demands new frameworks for oversight, ethics, and resilience in complex systems.