推理生命周期中的幻觉:大型推理模型中的接口可见性、因果证据与发布控制
Hallucination Across the Reasoning Lifecycle: Interface Visibility, Causal Evidence, and Release Control in Large Reasoning Models
浏览论文内容
中文总结 AI 辅助
本综述综合312篇论文,提出UIPCA框架分类推理幻觉,并指出尚无干预措施能在匹配条件下同时提升推理能力与事实可靠性,强调诊断、修复与发布控制。
中文摘要 AI 辅助
推理错误可能传播到后续决策和记忆中。本综述综合了312篇论文和第一方报告,围绕三个问题对基于文本的推理幻觉进行探讨:哪些证据是可观察的,哪些研究设计能确立因果关系,以及哪些纠正措施得到证据支持。UIPCA记录了无依据前提(U)、无效推理(I)、依赖重用(P)、可见答案轨迹一致性(C)和行动策略失败(A)。在58个被审查的来源中,没有比较研究证明在匹配条件下,某种特定干预能在提高推理能力的同时不降低事实可靠性。该综述将诊断与验证、修复、选择性发布以及跨记忆、工具和训练反馈的持久状态控制联系起来。
英文摘要
Reasoning errors can propagate into later decisions and memory. This survey synthesizes 312 papers and first-party reports on text-based reasoning hallucinations around three questions: what evidence is observable, what study designs establish, and which corrective actions the evidence supports. UIPCA records unsupported premises (U), invalid inferences (I), dependent reuse (P), visible answer-trace consistency (C), and action-policy failures (A). Across 58 reviewed sources, no comparison establishes that a specified intervention improves reasoning while reducing factual reliability under matched conditions. The synthesis connects diagnosis to verification, repair, selective release, and persistent-state control across memory, tools, and training feedback.
发表机构
- Zhejiang University(浙江大学)
- Binjiang Institute of Zhejiang University(浙江大学滨江研究院)
- Guangzhou University(广州大学)
- Rensselaer Polytechnic Institute(伦斯勒理工学院)
- The University of Hong Kong(香港大学)
- The Hong Kong University of Science and Technology (Guangzhou)(香港科技大学(广州))
- The Hong Kong Polytechnic University(香港理工大学)
- Zhejiang Lab(之江实验室)
机构由 AI 辅助整理,请以论文原文为准。