Strategy Adaptation in Large Language Model Werewolf Agents

TL;DR

This paper introduces a strategy adaptation method for Werewolf agents that dynamically switches strategies based on game context and player attitudes, improving performance over fixed or implicit strategies.

Contribution

It presents an explicit strategy selection approach for Werewolf agents that adapts to game dynamics, unlike prior fixed or implicit methods.

Findings

Strategy adaptation improves agent performance

Explicit context-based strategy selection outperforms fixed strategies

The method effectively estimates player roles and attitudes

Abstract

This study proposes a method to improve the performance of Werewolf agents by switching between predefined strategies based on the attitudes of other players and the context of conversations. While prior works of Werewolf agents using prompt engineering have employed methods where effective strategies are implicitly defined, they cannot adapt to changing situations. In this research, we propose a method that explicitly selects an appropriate strategy based on the game context and the estimated roles of other players. We compare the strategy adaptation Werewolf agents with baseline agents using implicit or fixed strategies and verify the effectiveness of our proposed method.