研究人员确定了大型语言模型中两种不同的弃答维度：答案正确性和问题可回答性。当前模型难以区分拒绝回答不正确的问题和拒绝回答根本不应回答的问题（例如，前提错误的提问）。一种使用隐藏状态探测而非简单置信度分数的新方法，可以更好地识别无法回答的问题，从而提高模型响应的准确性。 AI
影响 这项研究可能促使大型语言模型在拒绝回答不当或不正确的问题方面更加可靠，从而提高用户信任度和安全性。
排序理由 该集群包含两篇在 arXiv 上发表的关于大型语言模型行为和评估的学术论文。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
arXiv:2607.08456v1 Announce Type: cross Abstract: A model should refuse two different things: answers it would get wrong, and questions it should not answer at all, such as unanswerable ones or ones resting on a false premise. The usual recipe thresholds a single confidence score…
arXiv:2510.12857v2 Announce Type: replace-cross Abstract: Large language models (LLMs) are now widely deployed in user-facing applications, reaching hundreds of millions of users worldwide. Despite their widespread adoption, growing reliance on their outputs raises significant co…
A model should refuse two different things: answers it would get wrong, and questions it should not answer at all, such as unanswerable ones or ones resting on a false premise. The usual recipe thresholds a single confidence score, which cannot tell these apart. Across five instr…