# CAMEL: Communicative Agents for "Mind" Exploration of Large Scale Language Model Society

**Authors:** Guohao Li, Hasan Abed Al Kader Hammoud, Hani Itani, Dmitrii Khizbullin, Bernard Ghanem  
**arXiv:** [2303.17760](https://arxiv.org/abs/2303.17760)  
**Date:** 2023-03-31  
**Summary page:** https://www.summarizepaper.com/en/paper/2303.17760v1/

## Key points


## Summary

The field of artificial intelligence is rapidly advancing, and the success of AI systems heavily relies on the dynamics of interactions between agents. This paper focuses on building scalable techniques to facilitate autonomous cooperation among communicative agents and provide insight into their "cognitive" processes. To address the challenges associated with requesting chat agents to autonomously cooperate on completing tasks such as role flipping, assistant repeats instruction, flake replies, infinite loop of messages, and conversation termination conditions, the authors propose a novel cooperative agent framework named role-playing that allows communicative agents to collaborate autonomously towards task completion while requiring minimal human intervention. The proposed approach involves using inception prompting with role-playing to guide chat agents towards task completion while maintaining consistency with human intentions. The authors showcase how role-playing can be used to generate conversational data for studying the behaviors and capabilities of chat agents. They consider two cooperative scenarios of role-playing and generate two large conversational datasets: AI Society and Code. These datasets offer a valuable resource for investigating conversational language models, enabling them to comprehend and react to human language more effectively. The authors' contributions include introducing a novel communicative agent framework that offers a scalable approach for studying the cooperative behaviors and capabilities of multi-agent systems. They have also open-sourced their library containing implementations of various agents, data generation pipelines, data analysis tools, and collected datasets to support research on communicative agents and beyond. Overall, this paper provides insights into the challenges faced in achieving autonomous cooperation among communicative agents and proposes a promising solution through the use of role-playing with inception prompting. The generated datasets offer valuable resources for further investigation into conversational language models' behaviors and capabilities.