Do as We Do, Not as You Think: the Conformity of Large Language Models
Zhiyuan Weng, Guikun Chen, Wenguan Wang
TL;DR
This paper investigates conformity in large language model (LLM) driven multi-agent systems by introducing BenchForm, a conformity-centric benchmark derived from BIG-Bench Hard. It uses five Asch-inspired interaction protocols to quantify conformity via metrics $CR^P$ and $IR$, evaluating 11 LLMs and revealing widespread conformity influenced by interaction time and majority size. The study identifies protocol- and model-dependent variations, with the Doubt protocol most effective at inducing errors, and demonstrates mitigation possibilities through empowered personas and reflection/double-check prompts. The findings carry implications for the reliability and ethics of collaborative AI, and the authors provide BenchForm and code to enable future research in this area.
Abstract
Recent advancements in large language models (LLMs) revolutionize the field of intelligent agents, enabling collaborative multi-agent systems capable of tackling complex problems across various domains. However, the potential of conformity within these systems, analogous to phenomena like conformity bias and groupthink in human group dynamics, remains largely unexplored, raising concerns about their collective problem-solving capabilities and possible ethical implications. This paper presents a comprehensive study on conformity in LLM-driven multi-agent systems, focusing on three aspects: the existence of conformity, the factors influencing conformity, and potential mitigation strategies. In particular, we introduce BenchForm, a new conformity-oriented benchmark, featuring reasoning-intensive tasks and five distinct interaction protocols designed to probe LLMs' behavior in collaborative scenarios. Several representative LLMs are evaluated on BenchForm, using metrics such as conformity rate and independence rate to quantify conformity's impact. Our analysis delves into factors influencing conformity, including interaction time and majority size, and examines how the subject agent rationalizes its conforming behavior. Furthermore, we explore two strategies to mitigate conformity effects, i.e., developing enhanced personas and implementing a reflection mechanism. Several interesting findings regarding LLMs' conformity are derived from empirical results and case studies. We hope that these insights can pave the way for more robust and ethically-aligned collaborative AI systems. Our benchmark and code are available at BenchForm.