Long-Horizon Dialogue Understanding for Role Identification in the Game of Avalon with Large Language Models

TL;DR

This paper investigates the ability of large language models to understand long-horizon deception and role identification in the social deduction game Avalon, providing a new dataset and testbed for benchmarking LLMs in complex multi-party dialogues.

Contribution

The authors introduce a novel dataset and online testbed based on Avalon, enabling evaluation of LLMs' capabilities in long-horizon deception and role inference in multi-party interactions.

Findings

Current state-of-the-art LLMs underperform compared to humans in this task.

The dataset reveals the challenges LLMs face in understanding deception and motivation.

Multimodal integration of chat and game state offers insights into player identities.

Abstract

Deception and persuasion play a critical role in long-horizon dialogues between multiple parties, especially when the interests, goals, and motivations of the participants are not aligned. Such complex tasks pose challenges for current Large Language Models (LLM) as deception and persuasion can easily mislead them, especially in long-horizon multi-party dialogues. To this end, we explore the game of Avalon: The Resistance, a social deduction game in which players must determine each other's hidden identities to complete their team's objective. We introduce an online testbed and a dataset containing 20 carefully collected and labeled games among human players that exhibit long-horizon deception in a cooperative-competitive setting. We discuss the capabilities of LLMs to utilize deceptive long-horizon conversations between six human players to determine each player's goal and motivation. Particularly, we discuss the multimodal integration of the chat between the players and the game's state that grounds the conversation, providing further insights into the true player identities. We find that even current state-of-the-art LLMs do not reach human performance, making our dataset a compelling benchmark to investigate the decision-making and language-processing capabilities of LLMs. Our dataset and online testbed can be found at our project website: https://sstepput.github.io/Avalon-NLU/