Agent social dynamics improve teamwork in large language models
Agent Sociodrama: Social-Dynamics-Guided Persona Recasting for LLM Agent Teams
Artificial IntelligenceComputation and LanguageMultiagent Systems
Summary
Teams of large language model agents can struggle socially, like staying quiet or being overly dominant, leading to mistakes. The authors created Agent Sociodrama, a system that tracks and updates the feelings and relationships of these agents to better mimic real social interactions. They also developed a way to fix team problems by changing agents’ roles based on these social patterns. Their approach improved team performance significantly in tasks requiring cooperation and negotiation. This shows that understanding and managing agent social dynamics helps build better AI teams.
What this means in practice
- •For ai development teams: Design and test agent teams that adapt social roles to improve collaborative AI task success.
- •For game designers: Create believable AI characters whose social interactions evolve realistically during gameplay.
Authors
Yirui Qi, Xiaoming Zhang, Jaehong Yoon
Abstract
Large language model (LLM) agents are increasingly deployed as teams, yet they fail in characteristically social ways: information holders stay silent, dominant speakers anchor groups on wrong answers, and norm violations go unsanctioned. We present Agent Sociodrama, a framework for modeling and validating evolving social dynamics in LLM agent teams. Under fixed personas, agents' emotional states and directed, asymmetric relationships are updated every round by a unified external LLM judge and shape their subsequent interactions. Across six LLM backbones and five social scenarios, we assess social fidelity through classical social-theory patterns and agreement with human judgments. The resulting trajectories exhibit structured social patterns and model-specific differences. We then introduce Trajectory-Guided Persona Recasting (TPR), which diagnoses group-dynamic failures from interaction transcripts and social trajectories and recasts agent personas before rerunning the task. Across four collaborative benchmarks, TPR improves performance from both heterogeneous and randomly initialized agent persona compositions, yielding up to 83.3%p of accuracy improvement over the initial casting. On representative distributed-information cooperation and mixed-motive bargaining tasks, we outperform SOTOPIA and AgentVerse in three of four comparisons, by up to 37.5%p in Pareto rate. These results show that validated social trajectories provide actionable signals for repairing LLM agent teams.