The field of artificial intelligence is witnessing significant developments in the integration of multi-agent systems and large language models. Recent research has focused on enhancing the safety, efficiency, and decision-making capabilities of these systems. Notably, the use of large language models as mediators in multi-agent collaboration has shown promising results in medical decision-making and other applications. Furthermore, advancements in memory-augmented agents and multimodal systems have improved the ability of these systems to reason, learn, and interact with their environment. The development of novel frameworks and architectures, such as those leveraging active inference and probabilistic supernet sampling, has also contributed to the progress in this area. Overall, the field is moving towards more sophisticated, adaptive, and interpretable systems that can effectively collaborate and make decisions in complex scenarios. Noteworthy papers include: MedOrch, which proposes a mediator-guided multi-agent collaboration framework for medical decision-making, and PASS, which introduces a probabilistic agentic supernet sampling approach for interpretable and adaptive chest X-ray reasoning.
Advances in Multi-Agent Systems and Large Language Models
Sources
A Framework for Inherently Safer AGI through Language-Mediated Active Inference
Discovering Properties of Inflectional Morphology in Neural Emergent Communication
Mediator-Guided Multi-Agent Collaboration among Open-Source Models for Medical Decision-Making
Memp: Exploring Agent Procedural Memory
AquaChat++: LLM-Assisted Multi-ROV Inspection for Aquaculture Net Pens with Integrated Battery Management and Thruster Fault Tolerance
AdjustAR: AI-Driven In-Situ Adjustment of Site-Specific Augmented Reality Content
Kairos: Low-latency Multi-Agent Serving with Shared LLMs and Excessive Loads in the Public Cloud
EndoAgent: A Memory-Guided Reflective Agent for Intelligent Endoscopic Vision-to-Decision Reasoning
Triple-S: A Collaborative Multi-LLM Framework for Solving Long-Horizon Implicative Tasks in Robotics
Grounding Natural Language for Multi-agent Decision-Making with Multi-agentic LLMs
SimViews: An Interactive Multi-Agent System Simulating Visitor-to-Visitor Conversational Patterns to Present Diverse Perspectives of Artifacts in Virtual Museums
Multi-agent systems for chemical engineering: A review and perspective
SHIELDA: Structured Handling of Exceptions in LLM-Driven Agentic Workflows
WeChat-YATT: A Simple, Scalable and Balanced RLHF Trainer
TeamMedAgents: Enhancing Medical Decision-Making of LLMs Through Structured Teamwork
MinionsLLM: a Task-adaptive Framework For The Training and Control of Multi-Agent Systems Through Natural Language
CARES: Collaborative Agentic Reasoning for Error Detection in Surgery
Designing Memory-Augmented AR Agents for Spatiotemporal Reasoning in Personalized Task Assistance
Feedback-Driven Tool-Use Improvements in Large Language Models via Automated Build Environments
Intrinsic Memory Agents: Heterogeneous Multi-Agent LLM Systems through Structured Contextual Memory
Seeing, Listening, Remembering, and Reasoning: A Multimodal Agent with Long-Term Memory
Training-Free Multimodal Large Language Model Orchestration
FedCoT: Communication-Efficient Federated Reasoning Enhancement for Large Language Models
PASS: Probabilistic Agentic Supernet Sampling for Interpretable and Adaptive Chest X-Ray Reasoning
Reinforced Language Models for Sequential Decision Making