Modular Discovery of General Game-Playing Algorithms with Large Language Models

2026-09-27

AI Summary

This study addresses the challenge of designing search algorithms for general game playing, where game heterogeneity and strict decision-time constraints pose significant difficulties. We propose a large language model-based multi-agent meta-learning framework that automatically generates and refactors programmatic C++ search mechanisms and domain-specific heuristics through multi-agent co-evolution. This system discovers general game-playing algorithms without human intervention, employing AlphaRank for evolutionary evaluation. Experimental results demonstrate that our approach significantly outperforms fifteen Monte Carlo Tree Search baselines across over 400 game environments, exhibiting strong generalization capabilities and competitive advantages. Ultimately, this work establishes a fully automated algorithm discovery paradigm for solving general games.

Abstract

General Game Playing across arbitrary games from rules alone remains challenging due to differing algorithmic requirements across game classes and strict decision-time constraints. Rather than hand-designing search heuristics for specific domains, can we leverage Large Language Models (LLMs) to discover general game-playing algorithms? Because language models can propose and refactor structured code, they provide an expressive proposal engine for exploring the space of algorithmic designs. We introduce a multi-agent LLM meta-learning system to co-evolve game-agnostic procedural search mechanisms in C++ alongside domain heuristics synthesized directly from game rules. Controlling the compute budget, we benchmark the discovered mechanisms across more than 400 diverse environments, including OpenSpiel training and held-out games, procedural simulation engines, and games with deep neural policy-value representations trained via PPO. Evaluated via AlphaRank stationary distributions and Soft Condorcet Optimization (SCO) against 15 established MCTS baselines, the discovered search mechanisms consistently achieve top-tier ratings and pairwise ballot majorities over most baselines across independent evolutionary runs, generalizing to unseen human-designed and procedurally synthesized games and remaining competitive with baselines on frozen neural network representations.

Problem

• General Game Playing
• Large Language Models
• Algorithm Discovery
• Search Heuristics

Innovation

• Large Language Models
• General Game Playing
• Multi-agent Meta-learning
• Algorithm Discovery
• Monte Carlo Tree Search