Social Dynamics as Critical Vulnerabilities that Undermine Objective Decision-Making in LLM Collectives

Research examining how social dynamics create critical vulnerabilities in LLM collective systems. The paper explores mechanisms through which social influence can undermine objective decision-making when multiple language models interact. Contributes to growing research on LLM safety and alignment in collaborative multi-agent settings.