AI RESEARCH
The Cost of Consensus: Isolated Self-Correction Prevails Over Unguided Homogeneous Multi-Agent Debate
arXiv CS.AI
•
ArXi:2605.00914v1 Announce Type: cross Multi-agent debate, where teams of LLMs iteratively exchange rationales and vote on answers, is widely deployed under the assumption that peer review filters hallucinations. Yet the failure dynamics of homogeneous debate remain poorly understood,. therefore. we report findings from a controlled empirical study of teams of $N{=}10$ homogeneous agents (Qwen2.5-7B, Llama-3.1-8B, Ministral-3-8B) across $R{=}3$ debate rounds on two high-difficulty benchmarks (GSM-Hard and MMLU-Hard.