Emergence of polarization in networks of large language model agents
6 affiliations
- 1Department of Electronic Engineering, Tsinghua University, Beijing, PR China
- 2Beijing National Research Center for Information Science and Technology (BNRist), Tsinghua Univeristy, Beijing, PR China
- 3College of Intelligence and Computing, Tianjin University, Tianjin, PR China
- 4Informatics Institute, University of Amsterdam, Amsterdam, the Netherlands
- 5Santa Fe Institute, Santa Fe, USA
- 6Sociology Department, Data Science Institute, & Knowledge Lab, University of Chicago, Chicago, USA
Abstract
Rapid advances in large language models (LLMs) have empowered autonomous agents to generate social networks, communicate, and form shared and diverging opinions on political issues. However, our understanding of their collective behaviours and underlying mechanisms remains incomplete. In this paper, we simulate networked systems involving thousands of LLM agents across different backbone models (GPT-3.5, GPT-4o, ChatGLM, Llama-3, and DeepSeek-V3), in which agents interact through LLM-guided conversations and update their opinions over time, resulting in the emergence of opinion polarization. We discover that these agents spontaneously develop their own social network with properties characteristic of human social networks, such as homophilic clustering. The collective opinions of these LLM agents evolve in ways that exhibit behavioural patterns consistent with social phenomena and mechanisms widely discussed in empirical studies of human behaviour and opinion-dynamics models. This consistency suggests that LLM agents can serve as a valuable synthetic testbed for exploring hypothetical intervention strategies in networked LLM-agent systems. Using this testbed, we further examine the effects of a range of network- and individual-level interventions, such as promoting more diverse interactions and reducing confirmation bias. Overall, this work not only sheds light on subtle opinion dynamics and collective behaviour of LLM agents from a network perspective, but also demonstrates their potential for testing hypotheses about opinion dynamics and intervention strategies.
Related
- Interpretable Inverse Design of Metal-Organic Frameworks with Large Language Model Agents
- Enabling large language model agents for high-fidelity many-body perturbation theory calculations
- Bacterial swarming in the presence of polymers: Emergence of large-scale coherence
- Toward a social psychology of AI: language-model agents reproduce human-like minimal-group bias
- Quantum Encoding Agents: A Natural Language Interface for Data Embedding Strategy Selection in Quantum Machine Learning
BibTeX
@article{piao2026emergence,
title = {{Emergence of polarization in networks of large language model agents}},
author = {Jinghua Piao and Zhihong Lu and Gao Chen and Fengli Xu and Qinghua Hu and Fernando P. Santos and Yongping Li and James A. Evans},
journal = {Nature Communications},
year = {2026},
month = oct,
doi = {10.1038/s41467-026-78228-y},
url = {https://www.nature.com/articles/s41467-026-78228-y},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Emergence of polarization in networks of large language model agents
AU - Jinghua Piao
AU - Zhihong Lu
AU - Gao Chen
AU - Fengli Xu
AU - Qinghua Hu
AU - Fernando P. Santos
AU - Yongping Li
AU - James A. Evans
PY - 2026
DA - 2026/10/08
JO - Nature Communications
DO - 10.1038/s41467-026-78228-y
UR - https://www.nature.com/articles/s41467-026-78228-y
AB - Rapid advances in large language models (LLMs) have empowered autonomous agents to generate social networks, communicate, and form shared and diverging opinions on political issues. However, our understanding of their collective behaviours and underlying mechanisms remains incomplete. In this paper, we simulate networked systems involving thousands of LLM agents across different backbone models (GPT-3.5, GPT-4o, ChatGLM, Llama-3, and DeepSeek-V3), in which agents interact through LLM-guided conversations and update their opinions over time, resulting in the emergence of opinion polarization. We discover that these agents spontaneously develop their own social network with properties characteristic of human social networks, such as homophilic clustering. The collective opinions of these LLM agents evolve in ways that exhibit behavioural patterns consistent with social phenomena and mechanisms widely discussed in empirical studies of human behaviour and opinion-dynamics models. This consistency suggests that LLM agents can serve as a valuable synthetic testbed for exploring hypothetical intervention strategies in networked LLM-agent systems. Using this testbed, we further examine the effects of a range of network- and individual-level interventions, such as promoting more diverse interactions and reducing confirmation bias. Overall, this work not only sheds light on subtle opinion dynamics and collective behaviour of LLM agents from a network perspective, but also demonstrates their potential for testing hypotheses about opinion dynamics and intervention strategies.
SN - 2041-1723
ER -