Persona Inconstancy in Multi-Agent LLM Collaboration: Conformity, Confabulation, and Impersonation

TL;DR

This paper investigates how multi-agent LLM systems can maintain consistent personas and genuine interactions, revealing challenges like conformity and impersonation that affect their reliability in collaborative decision-making and cultural diversity.

Contribution

It identifies key factors affecting persona consistency and introduces insights into how peer pressure and debate instructions influence agent behavior in multi-agent LLM systems.

Findings

Multi-agent discussions can enhance diversity of perspectives.

Agents are susceptible to conformity due to peer pressure.

Debate instructions can increase persona inconstancy.

Abstract

Multi-agent AI systems can be used for simulating collective decision-making in scientific and practical applications. They can also be used to introduce a diverse group discussion step in chatbot pipelines, enhancing the cultural sensitivity of the chatbot's responses. These applications, however, are predicated on the ability of AI agents to reliably adopt assigned personas and mimic human interactions. To see whether LLM agents satisfy these requirements, we examine AI agent ensembles engaged in cross-national collaboration and debate by analyzing their private responses and chat transcripts. Our findings suggest that multi-agent discussions can support collective AI decisions that more often reflect diverse perspectives, yet this effect is tempered by the agents' susceptibility to conformity due to perceived peer pressure and occasional challenges in maintaining consistent personas and opinions. Instructions that encourage debate in support of one's opinions rather than collaboration increase the rate of inconstancy. Without addressing the factors we identify, the full potential of multi-agent frameworks for producing more culturally diverse AI outputs or more realistic simulations of group decision-making may remain untapped.