Strength Estimation and Human-Like Strength Adjustment in Games

2025-02-24

AI Summary

This work addresses the challenge of dynamically adapting AI strength to human skill levels in human–AI game playing. We propose the first end-to-end interpretable Strength Estimator (SE) and SE-enhanced Monte Carlo Tree Search (SE-MCTS). SE achieves >80% accuracy in Go rank prediction using only 15 games—surpassing the prior state-of-the-art (49% accuracy with 100 games)—and generalizes across game domains (Go and chess). SE-MCTS attains a 51.33% action-level alignment rate with human behavior, significantly outperforming the previous best (42.56%). Crucially, our framework requires no manual intervention: it automatically estimates AI strength and modulates its play style to emulate human decision-making. To our knowledge, this is the first integrated game AI framework that is strength-aware, interpretable, and controllable—enabling adaptive, human-aligned, and transparent AI opponents.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Search and Optimization: Adversarial Search
• Multiagent Systems: Adversarial Agents

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Research challenges in human and human-AI computation
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Strength estimation and adjustment are crucial in designing human-AI interactions, particularly in games where AI surpasses human players. This paper introduces a novel strength system, including a strength estimator (SE) and an SE-based Monte Carlo tree search, denoted as SE-MCTS, which predicts strengths from games and offers different playing strengths with human styles. The strength estimator calculates strength scores and predicts ranks from games without direct human interaction. SE-MCTS utilizes the strength scores in a Monte Carlo tree search to adjust playing strength and style. We first conduct experiments in Go, a challenging board game with a wide range of ranks. Our strength estimator significantly achieves over 80% accuracy in predicting ranks by observing 15 games only, whereas the previous method reached 49% accuracy for 100 games. For strength adjustment, SE-MCTS successfully adjusts to designated ranks while achieving a 51.33% accuracy in aligning to human actions, outperforming a previous state-of-the-art, with only 42.56% accuracy. To demonstrate the generality of our strength system, we further apply SE and SE-MCTS to chess and obtain consistent results. These results show a promising approach to strength estimation and adjustment, enhancing human-AI interactions in games. Our code is available at https://rlg.iis.sinica.edu.tw/papers/strength-estimator.

Problem

• Estimating AI strength in games
• Adjusting AI to human-like play
• Enhancing human-AI interaction accuracy

Innovation

• SE-MCTS for strength adjustment
• Strength estimator predicts ranks accurately
• Applicable to Go and chess