Exploring Large Language Models for Communication Games: An Empirical Study on Werewolf

TL;DR

This paper investigates how large language models can participate in communication games like Werewolf without tuning, using a retrieval and reflection-based framework, and observes emerging strategic behaviors.

Contribution

Proposes a tuning-free framework for LLMs to engage in communication games by leveraging retrieval and reflection, enabling effective play without parameter tuning.

Findings

LLMs can effectively play Werewolf without tuning.

Emergence of strategic behaviors in LLMs during the game.

Framework demonstrates potential for broader communication game applications.

Abstract

Communication games, which we refer to as incomplete information games that heavily depend on natural language communication, hold significant research value in fields such as economics, social science, and artificial intelligence. In this work, we explore the problem of how to engage large language models (LLMs) in communication games, and in response, propose a tuning-free framework. Our approach keeps LLMs frozen, and relies on the retrieval and reflection on past communications and experiences for improvement. An empirical study on the representative and widely-studied communication game, ``Werewolf'', demonstrates that our framework can effectively play Werewolf game without tuning the parameters of the LLMs. More importantly, strategic behaviors begin to emerge in our experiments, suggesting that it will be a fruitful journey to engage LLMs in communication games and associated domains.