CLFeb 26, 2025

Stay Focused: Problem Drift in Multi-Agent Debate

arXiv:2502.19559v212 citationsh-index: 14
Originality Incremental advance
AI Analysis

This addresses a key limitation in multi-agent debate for AI researchers, though it is incremental as it builds on existing methods to solve a specific bottleneck.

The paper tackles the issue of problem drift in multi-agent debate, where discussions deviate from the initial problem over multiple turns, harming performance on complex tasks; they propose DRIFTJudge and DRIFTPolicy to detect and mitigate this drift, improving task performance.

Multi-agent debate - multiple instances of large language models discussing problems in turn-based interaction - has shown promise for solving knowledge and reasoning tasks. However, these methods show limitations when solving complex problems that require longer reasoning chains. We analyze how multi-agent debate over multiple turns drifts away from the initial problem, thus harming task performance. We define this phenomenon as problem drift and quantify its presence across ten tasks (i.e., three generative, three knowledge, three reasoning, and one instruction-following task). To identify the reasons for this issue, eight human experts analyze 170 multi-agent discussions suffering from problem drift. We find the most common issues related to this drift are the lack of progress (35% of cases), low-quality feedback (26% of cases), and a lack of clarity (25% of cases). To address problem drift, we propose DRIFTJudge, an LLM-as-a-judge method, to detect problem drift at test-time. We also propose DRIFTPolicy, a method that mitigates problem drift cases to improve task performance. Our study is a step toward understanding a key limitation of multi-agent debate, highlighting why longer debates can harm task performance and how problem drift could be addressed.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes