arXiv cs.AI· Anthony Abou Mrad, Aviona Conti, Samuel Schmidgall, Jayanth Pratap, Ruhika Iyer, Joseph Kim, King Shi, Yann Sakref, Luis M. Ahumada, Lisa Knowlton, Andrew M. Cameron, Khyzer Aziz, Jeffrey K. Jopling·· 4 小时前AI 评分40
当评估变成干预:临床 AI 中 AI 评审模型的隐忧
When evaluation becomes intervention in clinical AI
AI 导读
一篇提交至 npj Digital Medicine 的论文探讨 AI 评审模型(AI judges,即按评分标准对其他系统输出打分的模型)在临床 AI 中的应用风险。当团队用评审结果来选择提示词或训练模型时,评审本身就成为干预的一部分,选择过程可能偏向评审所忽略的错误。论文提出了评审报告应包含的内容,以及每项由评审驱动的改动所需的独立临床测试。
正文 · 原文
Authors:Anthony Abou Mrad, Aviona Conti, Samuel Schmidgall, Jayanth Pratap, Ruhika Iyer, Joseph Kim, King Shi, Yann Sakref, Luis M. Ahumada, Lisa Knowlton, Andrew M. Cameron, Khyzer Aziz, Jeffrey K. Jopling
Abstract:AI judges, models that grade another system's output against a rubric, extend clinical review to the scale at which clinical AI writes. When teams use judge assessments to select a prompt or train a model, the judge becomes part of the intervention. Selection can then favor errors the judge overlooks. We describe what a judge's report must contain and the independent clinical test each judge-driven change needs.
| Comments: | 12 pages, 3 figures, 1 box. Submitted to npj Digital Medicine |
| Subjects: | Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computers and Society (cs.CY) |
| Cite as: | arXiv:2611.18297 [cs.AI] |
| (or arXiv:2611.18297v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.18297 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Anthony Abou Mrad [view email]
[v1]
Fri, 13 Nov 2026 15:21:50 UTC (739 KB)
来源:arXiv cs.AI · arxiv.org