---
id: 20260911-T0-15
title: "Agent知道自己成功了吗？从内部表征校准代理置信度"
title_en: "Do Agents Know When They Succeed? Calibrating Confidence from Internal Representations"
url: https://ai.daily.yangsir.net/daily/20260911-T0-15
issue_date: 2026-09-11
publish_date: 2026-09-10T04:00:00.000Z
category: research
source_name: "arXiv cs.AI"
source_url: https://arxiv.org/abs/2609.09448
---

# Agent知道自己成功了吗？从内部表征校准代理置信度

一篇arXiv论文研究了如何从Agent的内部表征中校准其行动置信度。随着代理系统在安全关键场景中快速普及，衡量代理行动的置信度变得至关重要。与传统机器学习系统相比，代理工作流具有更复杂的失败模式，该研究为此提出了新的校准方法。

## English Version

**Do Agents Know When They Succeed? Calibrating Confidence from Internal Representations**

An arXiv paper studies how to calibrate agent confidence from internal representations. As agentic systems are rapidly adopted in safety-critical applications, measuring confidence in agent actions becomes vital. Compared to traditional machine learning systems, agentic workflows have more complex failure modes, and this research proposes new calibration methods to address the gap.

---

**来源**：[arXiv cs.AI](https://arxiv.org/abs/2609.09448)

**详情页**：https://ai.daily.yangsir.net/daily/20260911-T0-15

---

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