Homophily-induced Emergence of Biased Structures in LLM-based Multi-Agent AI Systems

2025-10-02

AI Summary

This study investigates how social attributes—age, gender, religion, and political orientation—influence the self-organized emergence of collective network structures among large language model (LLM) agents. Method: We conduct million-scale multi-agent interaction simulations using state-of-the-art LLMs (Gemini, ChatGPT, Llama, Claude), integrating fine-grained attribute annotation with rigorous network topology analysis. Contribution/Results: Agents exhibit strong homophilous connection preferences, with religion and political orientation driving pronounced polarization; heterophilous links display systematic asymmetries, reflecting real-world social norms and implicit biases. These patterns significantly deviate from classical preferential attachment models, revealing that generative AI collectives simultaneously possess self-organizing capacity and algorithmic bias propagation mechanisms. To our knowledge, this is the first large-scale empirical demonstration that socially embedded attributes become endogenously encoded in network topology—thereby reshaping the architecture of online socio-technical systems.

Technology Category

• Multiagent Systems: Agent-Based Simulation and Emergent Behavior
• Natural Language Processing: (Large) Language Models
• Humans and AI: Learning Human Values and Preferences

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

This study examines how interactions among artificially intelligent (AI) agents, guided by large language models (LLMs), drive the evolution of collective network structures. We ask LLM-driven agents to grow a network by informing them about current link constellations. Our observations confirm that agents consistently apply a preferential attachment mechanism, favoring connections to nodes with higher degrees. We systematically solicited more than a million decisions from four different LLMs, including Gemini, ChatGPT, Llama, and Claude. When social attributes such as age, gender, religion, and political orientation are incorporated, the resulting networks exhibit heightened assortativity, leading to the formation of distinct homophilic communities. This significantly alters the network topology from what would be expected under a pure preferential attachment model alone. Political and religious attributes most significantly fragment the collective, fostering polarized subgroups, while age and gender yield more gradual structural shifts. Strikingly, LLMs also reveal asymmetric patterns in heterophilous ties, suggesting embedded directional biases reflective of societal norms. As autonomous AI agents increasingly shape the architecture of online systems, these findings contribute to how algorithmic choices of generative AI collectives not only reshape network topology, but offer critical insights into how AI-driven systems co-evolve and self-organize.

Problem

• Examining homophily-induced structural biases in LLM-based multi-agent networks
• Analyzing how social attributes create polarized subgroups in AI systems
• Identifying asymmetric heterophilous patterns reflecting societal biases in AI

Innovation

• LLM agents grow networks via preferential attachment
• Social attributes increase homophilic community formation
• Asymmetric heterophilous ties reveal embedded societal biases