Emergent social conventions and collective bias in LLM populations
arXiv:2410.08948 · doi:10.1126/sciadv.adu9368
Abstract
Social conventions are the backbone of social coordination, shaping how individuals form a group. As growing populations of artificial intelligence (AI) agents communicate through natural language, a fundamental question is whether they can bootstrap the foundations of a society. Here, we present experimental results that demonstrate the spontaneous emergence of universally adopted social conventions in decentralized populations of large language model (LLM) agents. We then show how strong collective biases can emerge during this process, even when agents exhibit no bias individually. Last, we examine how committed minority groups of adversarial LLM agents can drive social change by imposing alternative social conventions on the larger population. Our results show that AI systems can autonomously develop social conventions without explicit programming and have implications for designing AI systems that align, and remain aligned, with human values and societal goals.
References in corpus (12)
- Statistical physics of social dynamics
- A Survey on Large Language Model based Autonomous Agents
- Sharp transition towards shared vocabularies in multi-agent systems
- The Spontaneous Emergence of Conventions: An Experimental Study of Cultural Evolution
- Should ChatGPT be Biased? Challenges and Risks of Bias in Large Language Models
- Using Large Language Models to Simulate Multiple Humans and Replicate Human Subject Studies
- Machine Culture
- A new sociology of humans and machines
- Group interactions modulate critical mass dynamics in social convention
- CoAnnotating: Uncertainty-Guided Work Allocation between Human and Large Language Models for Data Annotation
- Shaping New Norms for AI
- Using Imperfect Surrogates for Downstream Inference: Design-based Supervised Learning for Social Science Applications of Large Language Models