Large Language Models can Achieve Social Balance

2024-10-05

arXiv.org

AI Summary

This work investigates whether large language model (LLM)-based multi-agent systems can spontaneously attain sociologically grounded “social balance”—i.e., convergence of relational networks into either a single cohesive group or stable opposing factions—during sustained interaction. Method: We design a multi-agent framework instantiated across three representative LLM families, integrating symbolic social network analysis, dynamic relational modeling, and cross-model comparative experiments to systematically assess the effects of interaction type, homophily preference, and group size on balance formation. Contribution/Results: We provide the first empirical evidence that LLM populations autonomously achieve social balance. Crucially, balance performance exhibits a non-monotonic relationship with model scale—largest models are not necessarily optimal. Quantitative analysis reveals significant inter-model variation in balance frequency, diversity of positive/negative relational patterns, and interaction stability, confirming distinct intrinsic socio-dynamic logics across architectures.

Technology Category

Machine Learning: Large Multimodal Models (LMMs)Multiagent Systems: Agent-Based Simulation and Emergent BehaviorCognitive Modeling & Cognitive Systems: Social Cognition And Interaction

Application Category

Social Networks and Social Media: Generative AI / large language models and their impact on social systemsSemantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactionsUser Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Social balance is a well-established concept in sociology which dictates how individual interactions can lead a population to become one faction of positive interactions or be divided in two or more antagonistic factions. In this paper, we consider a group of large language models (LLMs) and study how, after continuous interactions, they can achieve social balance. Across three different LLM models, we find that achieving social balance depends on (i) the type of interaction; (ii) whether agents consider homophily or influence from their peers; and (iii) the population size. We characterize how each model achieves social balance with different frequency, diversity of positive or negative interactions, and interaction stability across conditions (i) to (iii). We show that models achieve different notions of social balance and justify their social dynamics differently. Remarkably, the largest model is not necessarily more likely to achieve social balance with more frequency, stability, and diversity than the smaller ones.

Problem

• Study how LLMs achieve social balance through interactions.
• Examine factors like interaction type, homophily, and population size.
• Compare social balance frequency, stability, and diversity across LLMs.

Innovation

• LLMs achieve social balance through interactions.
• Social balance depends on interaction type, homophily, population size.
• Largest LLM not always better in social balance.