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

By Guohao Li, Hasan Hammoud, Hani Itani, Dmitrii Khizbullin, Bernard Ghanem. Neural Information Processing Systems, 2023

1,634 citations recorded by the source corpus.

The rapid advancement of chat-based language models has led to remarkable progress in complex task-solving. However, their success heavily relies on human input to guide the conversation, which can be challenging and time-consuming. This paper explores the potential of building scalable techniques to facilitate autonomous cooperation among communicative agents, and provides insight into their"cognitive"processes. To address the challenges of achieving autonomous cooperation, we propose a novel communicative agent framework named role-playing. Our approach involves using inception prompting to guide chat agents toward task completion while maintaining consistency with human intentions. We showcase how role-playing can be used to generate conversational data for studying the behaviors and capabilities of a society of agents, providing a valuable resource for investigating conversational language models. In particular, we conduct comprehensive studies on instruction-following cooperation in multi-agent settings. Our contributions include introducing a novel communicative agent framework, offering a scalable approach for studying the cooperative behaviors and capabilities of multi-agent systems, and open-sourcing our library to support research on communicative agents and beyond: https://github.com/camel-ai/camel.

View Paper: https://www.semanticscholar.org/paper/7bf72a3b5fbac8bc0f461780810fbc781c28ef53

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. Jeff Zhang
2. Jinyang Liu
3. Sai Qian Zhang
4. Chao Gao
5. Zeyu Han
6. Fanxu Meng

Citations in the corpus

• Improving Factuality and Reasoning in Language Models through Multiagent Debate (2023, unknown): “Largely concurrent to the work in this paper, a body of recent work has explored how multiagent interaction can be used with large language models (Zhang et al., 2023; Li et al., 2024; Hong et al., 2023).”
• Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing (2024, unknown): “The first type of method involves human effort to generate and curate instruction data”
• Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Survey (2024, background): “In the field of NLP, Large Language Models (LLMs) have achieved significant advancements across various tasks including text generation [10], [243], translation [239], [61], personalized chat-bots [192], [103], [187], and summarization [212], demonstrating remarkable proficiency.”
• Multi-Agent Collaboration Mechanisms: A Survey of LLMs (2025, unknown): “There are recent open-source frameworks allowing for experimentation with cooperative LLM-based MASs. CAMEL [74] provides a role-playing framework where a task-specific agent and two cooperating AI agents (User and Assistant) work to complete tasks via role-based conversations.”
• Empowering Biomedical Discovery with AI Agents (2024, unknown): “In such interdisciplinary systems, agents that could specialize in different biological domains, each with distinct capabilities, engage in interactions through various communication means.”
• Large language models empowered agent-based modeling and simulation: a survey and perspectives (2023, unknown): “CAMEL (Li et al., 2023d) introduces a cooperative role-playing framework with communicative agents, focusing on tool development.”
• What matters when building vision-language models? (2024, unknown): “Similarly to (Sanh et al., 2022; Wei et al., 2022; Bach et al., 2022; Dai et al., 2023; Li et al., 2023), each dataset is prompted into a shared question/answer format.”
• Federation of Agents: A Semantics-Aware Communication Fabric for Large-Scale Agentic AI (2025, unknown): “Dynamic orchestration remains partially solved: role-based systems improve structure yet often rely on manual wiring and do not couple decomposition with operational budgets [8, 9].”