---
id: 20260621-T0-05
title: "Hidden Anchors：多智能体LLM推理中的隐性偏见研究"
title_en: "Hidden Anchors in Multi-Agent LLM Deliberation"
url: https://ai.daily.yangsir.net/daily/20260621-T0-05
issue_date: 2026-06-21
publish_date: 2026-06-20T04:00:00.000Z
category: research
source_name: "arXiv cs.AI"
source_url: https://arxiv.org/abs/2606.19494
---

# Hidden Anchors：多智能体LLM推理中的隐性偏见研究

arXiv新论文分析了多智能体LLM协商中的"Hidden Anchors"现象。研究发现，尽管多轮对话能提高准确性，但也存在模仿人类决策的隐性偏差。

## English Version

**Hidden Anchors in Multi-Agent LLM Deliberation**

A new arXiv paper investigates "Hidden Anchors" in multi-agent LLM deliberation. It models how agents exchange answers and identifies hidden biases similar to human decision-making.

---

**来源**：[arXiv cs.AI](https://arxiv.org/abs/2606.19494)

**详情页**：https://ai.daily.yangsir.net/daily/20260621-T0-05

---

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