December 4, 2025PNAS NexusOpen Access

Network formation and dynamics among multi-LLMs

View Full Paper

Ask AI

Bookmark

LLMs demonstrate strong alignment in network formation with human social mobility patterns, indicating significant parallels in dynamics.
Key properties like homophily and preferential attachment emerge in LLM networks, showcasing structured community forms.
Analysis conducted involved benchmark comparisons across real-world social networks, emphasizing practical implications on fairness.
Insights highlight the potential role of LLMs in generating synthetic data while addressing risks of bias in interconnected environments.

Discussion

No takes yet. Share an insight, caveat, or question.

Observational analysis uncovered similarity in network formation behaviors of LLMs and humans, suggesting implications for fairness in AI systems.

Papachristou et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark

Observational analysis uncovered similarity in network formation behaviors of LLMs and humans, suggesting implications for fairness in AI systems.