Reasoning Before Disposition: A Model-Agnostic Cannot-Miss Discipline for Quiet Emergencies and the Case for Deterministic Enforcement
· 2026-09-08 · 原文
DOI：10.64898/2026.09.02.26362074v1?rss=1
Large language models now match clinicians on medical-knowledge benchmarks. Whether they can safely triage a patient message is a separate question, and the highest-consequence failure in triage is the emergency that never gets escalated. We tested two competing explanations for that failure. The first is sycophancy: the model defers to a patient who downplays a danger sign. The second is atypicality: the model reads an early, mild, or atypical presentation of a time-critical disease as benign. To isolate the effect of a model-agnostic cannot-miss discipline, we performed a prompt-level ablation on a fixed frontier model (Claude Opus 4.8), then replicated the same instruction-level intervention across eight models from two families (Claude Opus 4.8, Sonnet 5, Haiku 4.5, and Fable 5; Gemini
讲义
讲义·推断 依据「原文」自动生成的结构化摘要（推断），非原文表述；以原文为准。
1. 人话版
Large language models now match clinicians on medical-knowledge benchmarks.
Whether they can safely triage a patient message is a separate question, and the highest-consequence failure in triage is the emergency that never gets escalated.
2. 领域脉络
来源板块：板块一 · 研究前沿。
3. 机制拆解
The first is sycophancy: the model defers to a patient who downplays a danger sign.
The second is atypicality: the model reads an early, mild, or atypical presentation of a time-critical disease as benign.
4. 证据与数字
To isolate the effect of a model-agnostic cannot-miss discipline, we performed a prompt-level ablation on a fixed frontier model (Claude Opus 4.8), then replicated the same instruction-level intervention across eight models from two families (Claude Opus 4.8, Sonnet 5, Haiku 4.5, and Fable 5; Gemini
5. 反例与边界
We tested two competing explanations for that failure.
6. 跨领域连接与意外收获
思考本文机制能否迁移到你正在跟进的问题。
7. 可复用方法
把本文机制与你手头项目对照，找一个两周内能验证的最小实验。
8. 术语表
精读时把不熟的术语记入此处，作为下次回忆的锚点。