A survey on LLM-based multi-agent systems: workflow, infrastructure, and challenges
Обзор мультиагентных систем на основе больших языковых моделей: рабочий процесс, инфраструктура и проблемы
2024-10-08
SCID: 54.1/x9k6uzsj
Discuss with AI
LLM-based multi-agent systemsautonomous agentsgeneral artificial intelligencelarge language modelsworld simulation
Figures from the paper
Abstract (AI)
Abstract The pursuit of more intelligent and credible autonomous systems, akin to human society, has been a long-standing endeavor for humans. Leveraging the exceptional reasoning and planning capabilities of large language models (LLMs), LLM-based agents have been proposed and have achieved remarkable success across a wide array of tasks. Notably, LLM-based multi-agent systems (MAS) are considered a promising pathway towards realizing general artificial intelligence that is equivalent to or surpasses human-level intelligence. In this paper, we present a comprehensive survey of these studies, offering a systematic review of LLM-based MAS. Adhering to the workflow of LLM-based multi-agent systems, we synthesize a general structure encompassing five key components: profile, perception, self-action, mutual interaction, and evolution. This unified framework encapsulates much of the previous work in the field. Furthermore, we illuminate the extensive applications of LLM-based MAS in two principal areas: problem-solving and world simulation. Finally, we discuss in detail several contemporary challenges and provide insights into potential future directions in this domain.
Key Findings
1
It proposes a unified five-component workflow for LLM-based MAS: profile, perception, self-action, mutual interaction, and evolution.
2
LLM-based MAS applications are organized into two major domains: problem-solving and world simulation.
3
The framework synthesizes and organizes a substantial portion of prior LLM-based multi-agent research.
4
The survey identifies current challenges and outlines potential future research directions for LLM-based MAS development.
5
The survey systematically reviews research on large language model-based multi-agent systems (LLM-based MAS).
Research Object
LLM-based multi-agent systems
Research Subject
their workflow, infrastructure, applications, challenges, and future directions, including profile, perception, self-action, mutual interaction, and evolution
Publication Details
Publication Date
2024-10-08
Journal
Publisher
ISSN
Cited by
409
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai18
LLaMA: Open and Efficient Foundation Language Models2023
Overcoming catastrophic forgetting in neural networks2017
Large language models encode clinical knowledge2023
Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks2020
Training language models to follow instructions with human feedback2022
A survey on large language model based autonomous agents2024
HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units2021
Generative Agents: Interactive Simulacra of Human Behavior2023
PaLM: Scaling Language Modeling with Pathways2022
Flamingo: a Visual Language Model for Few-Shot Learning2022
AST: Audio Spectrogram Transformer2021
Autonomous chemical research with large language models2023
Active Retrieval Augmented Generation2023
Augmenting large language models with chemistry tools2024
ChatDev: Communicative Agents for Software Development2024
ExpeL: LLM Agents Are Experiential Learners2024
AgentTuning: Enabling Generalized Agent Abilities for LLMs2024
Enhancing Retrieval-Augmented Large Language Models with Iterative Retrieval-Generation Synergy2023
Cited by11
Improving large language model applications in biomedicine with retrieval-augmented generation: a systematic review, meta-analysis, and clinical development guidelines2025
The Rise of Agentic AI: A Review of Definitions, Frameworks, Architectures, Applications, Evaluation Metrics, and Challenges2025
Evaluation and Benchmarking of LLM Agents: A Survey2025
Generative AI and LLMs for Critical Infrastructure Protection: Evaluation Benchmarks, Agentic AI, Challenges, and Opportunities2025
Multi-agent systems powered by large language models: applications in swarm intelligence2025
Agentic AI in Healthcare and Medicine: A Seven-Dimensional Taxonomy for Empirical Evaluation of LLM-Based Agents2026
Multi-agent AI2026
A review on integrated machine learning and deep learning driven artificial intelligence models for pharmacokinetics and toxicokinetics predictions, and their application2026
ResQConnect: An AI-Powered Multi-Agentic Platform for Human-Centered and Resilient Disaster Response2026
Evaluating routing stability and coordination in swarm-based multi-agent task-oriented dialogue systems2026
Technical Limitations of LLM-Based AI Agents and Their Links to Bias and Governance Challenges: A Narrative Review2026