Multi-Agent LLMs Fail to Explore Each Other

Hyeong Kyu Choi Jiatong Li Wendi Li X. Wang Sharon Li

Jul 2026 · arXiv.org · Vol abs/2607.11250 · 1 citation

Computer Science

TL;DR

This work introduces Multi- Agent Contextual Exploration (MACE), a lightweight framework that explicitly promotes exploration through structured peer selection that substantially improves exploration behavior and downstream task performance and shows theoretically that the value of exploration increases with agent diversity.

Abstract

Exploration is essential for reliable autonomy in multi-agent systems, yet it remains unclear whether large language model (LLM) agents can explore effectively when interacting with one another. We show that modern LLM agents fail to do so, often exhibiting myopic and polarized interaction patterns that lead to suboptimal coordination and increased regret. We formalize this challenge as the Multi-Agent Exploration problem, modeling it as a partially observable stochastic game (POSG) problem in which agents must probe peers to infer their capabilities and identify effective interaction strategies. To address this, we introduce Multi- Agent Contextual Exploration (MACE), a lightweight framework that explicitly promotes exploration through structured peer selection. Across both contextual and parametric diversity settings, MACE substantially improves exploration behavior and downstream task performance. We further show theoretically that the value of exploration increases with agent diversity. Overall, our results highlight a fundamental limitation of current LLM agents and underscore the importance of explicitly guided exploration for reliable multi-agent autonomy. Code will be released in https://github.com/deeplearning-wisc/mace

DOI arXiv Semantic Scholar

View source

Similar papers

Jul 2026

When Do Multi-Agent Systems Help? An Information Bottleneck Perspective

This study shows that multi-agent design is fundamentally an information-bottleneck optimization problem, and explains when bounded inter-agent communication helps or hurts, and observes that MAS consistently helps when relays are near-sufficient, especially for weaker models.

Wendi Yu, Lianhao Zhou, Xiangjue Dong et al. · arXiv.org · 0 citations

Preprint Aug 2026

History Matters: Meta-policy Delegation with Heterogeneous Multi-agent Reinforcement Learning

This paper develops a multi-agent reinforcement learning-based (MARL) delegation training that enables agents to make sequential delegation decisions while minimizing the total execution cost and introduces two new frameworks for collaboration and delegation in multi-agent systems.

Ziqing Lu, Avinash Mudireddy, Sarra M. Alqahtani et al. · 0 citations

Conference Open access Jul 2026

AgentDropout: Dynamic Redundancy Elimination for Multi-Agent Collaboration Efficiency

AgentDropout is a dynamic strategy inspired by dropout regularization in neural networks, which selectively deactivates low-contribution agents during multi-agent collaboration, suggesting that dynamic agent deactivation may be useful for deploying multi-agent LLM systems under computational budget constraints.

Zhengxi Xiao, Qi Guo, Yuyue Wang et al. · 2026 8th International Confe... · 1 citation

Preprint Aug 2026

Why Study Emergent Behavior When You Can Regulate It? Aligning Multi-Agent Systems with Reward Prediction

The introduction of Multi-Agent Reward Prediction (MARP) demonstrates that emergent multi-agent behavior can be treated not only as a phenomenon to study, but as a target of principled, data-driven regulation.

Assaf Caftory, Almog Zemach, Moshe Butman et al. · 1 citation

Preprint Aug 2026

Training Small LLMs as Spatial Multi-Agent Policies

Training LLM-based multi-agent systems with multi-agent reinforcement learning with multi-agent reinforcement learning is rapidly gaining traction, and a parallel line of work argues that such systems should be judged by their behavior, not only their reward.

Yi Mao, Andrew Perrault · 0 citations

Preprint Aug 2026

Emergence of Biased Consensus in Multi-Agent LLM Debates

This work identifies the emergence of collective norms in multi-agent LLM debates and finds that agent heterogeneity suppresses emergence by smoothing this transition, and shows that these insights generalize to realistic decision-making tasks, including investment decisions and LLM-as-a-judge evaluation.

Maya Okawa · 1 citation