On the Dynamics of Multi-Agent LLM Communities Driven by Value Diversity

2025/12/11 by Mu‐Hua Huang, Huang, Muhua, Qinlin Zhao +5
Neuroscience · Social Sciences · #Artificial Intelligence (cs.AI) #Embodied and Extended Cognition #Ethics and Social Impacts of AI #FOS: Computer and information sciences #Language and cultural evolution

paper · pdf · doi:10.48550/arxiv.2512.10665

openalex publication_date 2025/12/11 · openalex created_date 2025/12/13 · openalex updated_date 2026/07/28

Abstract

As Large Language Models (LLM) based multi-agent systems become increasingly prevalent, the collective behaviors, e.g., collective intelligence, of such artificial communities have drawn growing attention. This work aims to answer a fundamental question: How does diversity of values shape the collective behavior of AI communities? Using naturalistic value elicitation grounded in the prevalent Schwartz's Theory of Basic Human Values, we constructed multi-agent simulations where communities with varying numbers of agents engaged in open-ended interactions and constitution formation. The results show that value diversity enhances value stability, fosters emergent behaviors, and brings more creative principles developed by the agents themselves without external guidance. However, these effects also show diminishing returns: extreme heterogeneity induces instability. This work positions value diversity as a new axis of future AI capability, bridging AI ability and sociological studies of institutional emergence.

Citations

• Evolving Collective Cognition in Human-Agent Hybrid Societies: How Agents Form Stances and Boundaries
• IROTE: Human-like Traits Elicitation of Large Language Model via In-Context Self-Reflective Optimization
• AdAEM: An Adaptively and Automated Extensible Measurement of LLMs' Value Difference
• DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning
• A systematic review of norm emergence in multi-agent systems
• Designing AI-Agents with Personalities: A Psychometric Approach
• Normative Modules: A Generative Agent Architecture for Learning Norms that Supports Multi-Agent Cooperation
• Organizing a Society of Language Models: Structures and Mechanisms for Enhanced Collective Intelligence
• Evolving AI Collectives to Enhance Human Diversity and Enable Self-Regulation
• Value FULCRA: Mapping Large Language Models to the Multidimensional Spectrum of Basic Human Values
• Direct Preference Optimization: Your Language Model is Secretly a Reward Model
• Sparks of Artificial General Intelligence: Early experiments with GPT-4
• Affective Coherence Monitoring for Transformer-Based Language Models
• Scaling Laws for Reward Model Overoptimization
• Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
• Training Compute-Optimal Large Language Models
• A General Language Assistant as a Laboratory for Alignment
• Scaling Laws for Neural Language Models

Related

• Cultural Evolution of Cooperation among LLM Agents
• Evolving AI Collectives to Enhance Human Diversity and Enable Self-Regulation
• Emergent social conventions and collective bias in LLM populations
• Increasing intelligence in AI agents can worsen collective outcomes
• The Role of Social Learning and Collective Norm Formation in Fostering Cooperation in LLM Multi-Agent Systems