---
id: 20260610-T0-04
title: "CARTOGRAPH：AI科学家实验验证层，能主动拒绝不可行任务"
title_en: "CARTOGRAPH: AI Scientist Verification with Active Refusal"
url: https://ai.daily.yangsir.net/daily/20260610-T0-04
issue_date: 2026-06-10
publish_date: 2026-06-09T04:00:00.000Z
category: research
source_name: "arXiv cs.LG (ML)"
source_url: https://arxiv.org/abs/2606.07576
---

# CARTOGRAPH：AI科学家实验验证层，能主动拒绝不可行任务

arXiv新研究提出CARTOGRAPH系统，专为AI科学家设计。该系统包含三核心功能：实验选择（在未解空间中定向）、歧义消除（明确处理模糊问题）、以及任务拒绝（检测不可行任务）。通过局部线性高斯假设验证，确保AI科学家只执行可验证的实验。此研究解决了AI自主研究中实验可靠性问题，未来可应用于复杂科学发现场景。

## English Version

**CARTOGRAPH: AI Scientist Verification with Active Refusal**

Researchers present CARTOGRAPH, a verification layer for AI scientists. It combines three functions: experiment steering in unresolved subspaces, explicit ambiguity closure, and task refusal for infeasible experiments. Under local linear-Gaussian assumptions, the system ensures AI only performs verifiable experiments, addressing reliability in autonomous scientific discovery.

---

**来源**：[arXiv cs.LG (ML)](https://arxiv.org/abs/2606.07576)

**详情页**：https://ai.daily.yangsir.net/daily/20260610-T0-04

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*