English(EN) Do Large Language Models Know What They Don't Know II? A Fully Behavioral, Non-Cognitive Measure of Epistemic Honesty

新的EHQ基准衡量LLM的情感诚实度，揭示知识差距

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-10 04:00

一篇新的研究论文介绍了一种新颖的衡量标准——情感诚实度商（EHQ），旨在评估大型语言模型（LLM）在多大程度上承认其知识的局限性。该研究构建了一个包含3000个问题的基准测试EHQ-3000，涵盖了虚构实体、截止日期后的事件、小众知识和条件性问题。对14个LLM API接口的分析显示，情感诚实度存在显著差异，综合EHQ得分范围从0.31到0.81不等，表明这种基于行为的衡量标准能够捕捉到标准正确性评估中不明显的差异。 AI

影响强调了更好地评估LLM的置信度和知识边界的必要性，可能影响未来的模型开发和安全测试。

排序理由介绍新基准和衡量标准以评估LLM行为的研究论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 17 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 介绍新基准和衡量标准以评估LLM行为的研究论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, safety; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Ali \c{S}enol, H. Russell Bernard, Huan Liu · 2026-09-10 04:00

大型语言模型是否知道自己不知道什么 II？一种完全行为化、非认知性的认知诚实度衡量方法

arXiv:2609.07879v1 Announce Type: new Abstract: Large Language Models (LLMs) are frequently confident, eloquent, and well versed. A natural question arises: do they know what they don't know? To answer this question, we borrow the concept of epistemic honesty and develop a novel …

本文如何被排名

报道来源 [1]

大型语言模型是否知道自己不知道什么 II？一种完全行为化、非认知性的认知诚实度衡量方法

相关实体

相关话题