Rethinking the Bounds of LLM Reasoning: Are Multi-Agent Discussions the Key?
This work questions the necessity of multi-agent systems for LLM reasoning, potentially simplifying approaches for AI researchers and practitioners.
The paper challenges the claim that multi-agent discussions improve LLM reasoning, showing that a single agent with strong prompts matches the best existing discussion approach across tasks and models, with multi-agent only outperforming when prompts lack demonstrations.
Recent progress in LLMs discussion suggests that multi-agent discussion improves the reasoning abilities of LLMs. In this work, we reevaluate this claim through systematic experiments, where we propose a novel group discussion framework to enrich the set of discussion mechanisms. Interestingly, our results show that a single-agent LLM with strong prompts can achieve almost the same performance as the best existing discussion approach on a wide range of reasoning tasks and backbone LLMs. We observe that the multi-agent discussion performs better than a single agent only when there is no demonstration in the prompt. Further study reveals the common interaction mechanisms of LLMs during the discussion.