多智能体卧底博弈：通过反事实测试消除多模态推理中的幻觉
Multi-agent Undercover Gaming: Hallucination Removal via Counterfactual Test for Multimodal Reasoning
AI总结：
针对大型语言模型多智能体辩论依赖理性假设、易因幻觉失效的问题，提出受社交推理游戏启发的多智能体卧底博弈协议，通过反事实测试检测幻觉智能体，提升多模态推理可靠性与有效性。
摘要：
幻觉仍然是大型语言模型（LLMs）推理能力的主要障碍。尽管多智能体辩论（MAD）范式通过促进多个智能体之间的共识来提高可靠性，提供了一个有前景的解决方案，但它依赖于所有辩论者都是理性且善于反思的不切实际假设，而当智能体自身容易产生幻觉时，这一条件可能无法成立。为解决这一差距，我们受“谁是卧底”等社交推理游戏启发，引入了多智能体卧底博弈（MUG）协议。MUG将MAD重构为通过多模态反事实测试检测“卧底”智能体（即产生幻觉的智能体）的过程。具体而言，我们修改参考图像以引入反事实证据，并观察智能体是否能准确识别这些变化，从而为识别幻觉智能体提供真实依据，实现稳健的群体驱动多模态推理。MUG从三个关键维度推进MAD协议：（1）通过反事实测试实现超越统计共识的事实验证；（2）通过动态修改的证据源引入跨证据推理，而非依赖静态输入；（3）促进主动推理，让智能体参与探究性讨论而非被动回答问题。这些创新共同为LLMs的多模态推理提供了更可靠、更有效的框架。源代码可通过https://github.com/YongLD/MUG.git获取。
英文摘要：
Hallucination continues to pose a major obstacle in the reasoning capabilities of large language models (LLMs). Although the Multi-Agent Debate (MAD) paradigm offers a promising solution by promoting consensus among multiple agents to enhance reliability, it relies on the unrealistic assumption that all debaters are rational and reflective, which is a condition that may not hold when agents themselves are prone to hallucinations. To address this gap, we introduce the Multi-agent Undercover Gaming (MUG) protocol, inspired by social deduction games like "Who is Undercover?". MUG reframes MAD as a process of detecting "undercover" agents (those suffering from hallucinations) by employing multimodal counterfactual tests. Specifically, we modify reference images to introduce counterfactual evidence and observe whether agents can accurately identify these changes, providing ground-truth for identifying hallucinating agents and enabling robust, crowd-powered multimodal reasoning. MUG advances MAD protocols along three key dimensions: (1) enabling factual verification beyond statistical consensus through counterfactual testing; (2) introducing cross-evidence reasoning via dynamically modified evidence sources instead of relying on static inputs; and (3) fostering active reasoning, where agents engage in probing discussions rather than passively answering questions. Collectively, these innovations offer a more reliable and effective framework for multimodal reasoning in LLMs. The source code can be accessed at https://github.com/YongLD/MUG.git.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
