핵심 개념
LLM-based agents exhibit adaptive social behaviors in Avalon gameplay, showcasing potential in dynamic interactions.
초록
The paper explores LLM-based agents' social behaviors in Avalon gameplay.
Proposed a novel framework for efficient communication and interaction.
Evaluated the framework's effectiveness in creating adaptive agents.
Analyzed social behaviors like teamwork, leadership, persuasion, deception, and confrontation.
Conducted experiments to validate the framework's performance.
Identified the importance of key modules in achieving winning strategies.
통계
"Our method demonstrated a 90% winning rate in 10 games when playing the good side."
"When playing the evil side, the winning rate was 100% over the same number of games."
"Our proposed Avalon Agents achieve a 100% win rate against the baseline agents when playing as the evil side."
인용구
"Our contributions can be summarized as: exploring social behaviors of LLM-based agents, designing an effective framework for Avalon, and shedding light on the implications of these behaviors."
"Our findings have the potential to contribute to a better understanding of the role of LLM-based agents in social and strategic contexts."