An Implementation of Werewolf Agent That does not Truly Trust LLMs

TL;DR

This paper presents a hybrid werewolf game agent combining LLMs and rule-based methods to improve conversational consistency, persona, and logical behavior, with qualitative evaluation showing increased human-likeness.

Contribution

The paper introduces a novel hybrid approach that integrates rule-based algorithms with LLMs to address challenges in werewolf game agents, enhancing consistency and persona.

Findings

Agent perceived as more human-like than unmodified LLM

Hybrid approach improves logical and situational responses

Agent can refute, end conversations, and behave with persona

Abstract

Werewolf is an incomplete information game, which has several challenges when creating a computer agent as a player given the lack of understanding of the situation and individuality of utterance (e.g., computer agents are not capable of characterful utterance or situational lying). We propose a werewolf agent that solves some of those difficulties by combining a Large Language Model (LLM) and a rule-based algorithm. In particular, our agent uses a rule-based algorithm to select an output either from an LLM or a template prepared beforehand based on the results of analyzing conversation history using an LLM. It allows the agent to refute in specific situations, identify when to end the conversation, and behave with persona. This approach mitigated conversational inconsistencies and facilitated logical utterance as a result. We also conducted a qualitative evaluation, which resulted in our agent being perceived as more human-like compared to an unmodified LLM. The agent is freely available for contributing to advance the research in the field of Werewolf game.