The emergence of social norms has attracted much interest in a wide array of disciplines, ranging from social science and cognitive science to artificial intelligence. In this paper, we propose the first generative agent architecture that empowers the emergence of social norms within a population of large language model-based agents. Our architecture, named CRSEC, consists of four modules: Creation & Representation, Spreading, Evaluation, and Compliance. Our architecture addresses several important aspects of the emergent processes all in one: (i) where social norms come from, (ii) how they are formally represented, (iii) how they spread through agents' communications and observations, (iv) how they are examined with a sanity check and synthesized in the long term, and (v) how they are incorporated into agents' planning and actions. Our experiments deployed in the Smallville sandbox game environment demonstrate the capability of our architecture to establish social norms and reduce social conflicts within large language model-based multi-agent systems. The positive outcomes of our human evaluation, conducted with 30 evaluators, further affirm the effectiveness of our approach.

翻译：社会规范的出现引起了从社会科学、认知科学到人工智能等多个学科领域的广泛关注。本文提出了首个生成式智能体架构，该架构能够在大语言模型智能体群体中实现社会规范的自发涌现。我们提出的架构名为CRSEC，包含四个模块：创建与表征、传播、评估与遵从。该架构统一处理了涌现过程的几个关键方面：(i) 社会规范的来源，(ii) 其形式化表征方式，(iii) 通过智能体通信与观察实现的传播机制，(iv) 长期范围内的合理性检验与综合合成，以及(v) 社会规范对智能体规划与行动的约束整合。我们在Smallville沙盒游戏环境中开展的实验表明，该架构能够在大语言模型驱动的多智能体系统中建立社会规范并减少社会冲突。由30名评估者参与的人工评估取得积极结果，进一步验证了本方法的有效性。