arXiv cs.CY· Mohsen Amoei, Kathleen Fraser, Dan Poenaru·· 13 小时前AI 评分31
临床医生与 LLM 的错误重叠：审计与治理互补性脆弱点
Complementary Vulnerabilities: Auditing and Governing Clinician-LLM Error Overlap
AI 导读
研究提出临床医生与 LLM 的"互补性脆弱点"概念：两者失效机制不同（医生是具身与认知层面，模型是统计与生成层面），因此第二道自动检查只有在与第一道以不同方式出错时才能降低错误。作者据此建议各机构在部署前用自身病例审计错误重叠，并建立治理机制长期维持错误独立性，评估临床 AI 工具时应同时考察准确率与失败重叠度。
正文 · 原文
Abstract:Health systems are deploying large language models (LLMs) into clinical workflows, often on the assumption that a second automated check can only reduce error. But a second reasoner reduces error only insofar as it fails differently from the first. We say that a clinician and a model have complementary vulnerabilities to the extent that they break down through different mechanisms and on different cases. The two reasoners' failure substrates differ: they are embodied and cognitive in the clinician, and statistical and generative in the model - so their weak points need not line up. That divergence is plausible rather than guaranteed, and the independence it creates is fragile, eroding when clinicians defer to the model or the model conforms to the clinician. We therefore propose a predeployment audit in which each site measures error overlap on its own cases. We also propose a governance model with which the site keeps those errors independent over time. Health systems should then judge a clinical artificial intelligence (AI) tool by whether its failures overlap with their clinicians' failures, as well as by its accuracy.
| Comments: | 4 pages main text, 2 figures. Accepted at SMASH 2026, the Symposium on Model Accountability, Sustainability and Healthcare (Mila, Montreal, November 3-4, 2026), after peer review on OpenReview |
| Subjects: | Computers and Society (cs.CY) |
| Cite as: | arXiv:2610.25279 [cs.CY] |
| (or arXiv:2610.25279v2 [cs.CY] for this version) | |
| https://doi.org/10.48550/arXiv.2610.25279 arXiv-issued DOI via DataCite |
Submission history
From: Mohsen Amoei Mr [view email]
[v1]
Sun, 18 Oct 2026 13:55:57 UTC (30 KB)
[v2]
Thu, 22 Oct 2026 01:14:40 UTC (30 KB)
来源：arXiv cs.CY · arxiv.org