# 斯坦福牛津 MedRSI：医疗智能体自我改进需护栏

- 来源：Rohan Paul (@rohanpaul_ai)
- 发布时间：2026-09-23 11:27
- AIHOT 分数：45
- AIHOT 链接：https://aihot.news/items/cmudk1zuy0draroggjmmk9pqt
- 原文链接：https://x.com/rohanpaul_ai/status/2102600637922124145

## AI 摘要

斯坦福与牛津论文 MedRSI 显示，医疗智能体可从自身错误中自我改进，但新能力须先在后续新患者上验证才能固化。若每轮都立即注册新工具，第 30 轮准确率降至 76.9%（57 个工具）；放慢注册、仅保留 18 个工具时准确率达 94.4%。该智能体还会优先处理可能造成更大临床危害的错误，而非仅关注最常见错误。

## 正文

New Stanford+Oxford paper MedRSI shows that medical agents can improve themselves from their own mistakes, but only if new capabilities are tested on fresh patients before becoming permanent.

shows self-improving agents need 2 things: focus on harmful failures and refuse to keep new capabilities until they work on later cases.

More important, the paper shows why self-improvement needs guardrails.

If every promising tool is added immediately, accuracy eventually falls: 76.9% by round 30 with 57 tools.

With slower registration, the agent kept just 18 tools and held 94.4%.

It also spends more effort on mistakes that could cause greater clinical harm, rather than just the most common errors.

let agents invent aggressively, but make permanent self-changes earn their place through repeated independent evaluation.