用生成式智能体模型探索因果机制
原标题：Exploring Causal Mechanisms with Generative Agent-Based Models
AI 摘要
该论文提出 RePair 方法，利用生成式智能体模型（LLM 智能体）探索因果机制：先校准模拟世界，再将候选机制表述为自然语言规则，通过匹配干预估计其对集体结果的影响，并检查行为轨迹。作者在居住隔离、公共品合作、群体极化与合作级联四个基于社会科学模型的模拟世界中测试该方法，发现自然语言规则能产生可测量的集体效应，规则比较随配置积累而收敛，行为轨迹可将集体效应与智能体行为联系起来。
正文节选
Exploring Causal Mechanisms with Generative Agent-Based Models Abstract In this paper, we explore using generative agent-based models for a classical ABM application: testing how individual-level behavioral rules produce collective phenomena. We introduce RePair, a method that calibrates simulation worlds, operationalizes candidate mechanisms as natural-language rules, estimates their effects through matched interventions, and examines behavioral traces. We assess the method by testing it in fou