paper-with-me

홈 › Papers

LLM Abstention Can Be a Prompt Artifact, in Addition to Genuine Uncertainty

2025-07-22 · Zipeng Ling, Shuliang Liu, Yuehao Tang, Junqi Yang, Shenghong Fu, Chen Huang, Kejia Huang, Yao Wan, Zhichao Hou, Xuming Hu arxiv

Large Language Models (LLMs) are increasingly trained to abstain from answering questions they are unsure about. However, this ability is often misused: in real-world applications, input prompts sometimes contain uncertainty elements, and driven by this, LLMs are inclined to abstain even on problems they are capable of solving. We argue that LLM abstention is not only an expression of genuine uncertainty; it is also an artifact that can be largely influenced by prompts. We name this phenomenon *Abstention Inflation*. We add "Unknown" as an extra option for LLMs to choose from; experiments show serious accuracy drops on True/False Questions (TFQs). Replacing "Unknown" with an unrelated random word produces an identical effect. We argue that LLMs are trained to imitate the surface pattern of *abstention*, rather than to express genuine uncertainty. Based on ten experiments, we support four claims that form a progressive argument: (C1) *Abstention Inflation* is triggered by the structural presence of an extra option, not by genuine uncertainty; (C2) further, it makes the model deny it can answer even when it can; (C3) at the representation level, this manifests as a later-layer output override; (C4) finally, this bias is stable and emerges through instruction tuning, rather than stochastic noise.

📄 PDF Abstract BibTeX arXiv:2507.16199

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Mind the Gap: How Elicitation Protocols Shape the Stated-Revealed Preference Gap in Language Models

2026-01-29 · Pranav Mahajan, Ihor Kendiukhov, Syed Hussain, Lydia Nottingham arxiv

Recent work identifies a stated-revealed (SvR) preference gap in language models (LMs): a mismatch between the values models endorse and the choices they make in context. Existing evaluations rely heavily on binary force…

Confidential Guardian: Cryptographically Prohibiting the Abuse of Model Abstention

2025-05-29 · Stephan Rabanser, Ali Shahin Shamsabadi, Olive Franzese, Xiao Wang 외

Cautious predictions -- where a machine learning model abstains when uncertain -- are crucial for limiting harmful errors in safety-critical applications. In this work, we identify a novel threat: a dishonest institution…

I-CALM: Incentivizing Confidence-Aware Abstention for LLM Hallucination Mitigation

2026-04-05 · Haotian Zong, Binze Li, Yufei Long, Sinyin Chang 외 arxiv

Large language models (LLMs) frequently produce confident but incorrect answers, partly because common binary scoring conventions reward answering over honestly expressing uncertainty. We study whether prompt-only interv…

AbstentionBench: Reasoning LLMs Fail on Unanswerable Questions

2025-06-10 · Polina Kirichenko, Mark Ibrahim, Kamalika Chaudhuri, Samuel J. Bell

For Large Language Models (LLMs) to be reliably deployed in both everyday and high-stakes domains, knowing when not to answer is equally critical as answering correctly. Real-world user queries, which can be underspecifi…

Math

Knowing When to Abstain: Medical LLMs Under Clinical Uncertainty

2026-01-18 · Sravanthi Machcha, Sushrita Yerra, Sahil Gupta, Aishwarya Sahoo 외 arxiv

Current evaluation of large language models (LLMs) overwhelmingly prioritizes accuracy; however, in real-world and safety-critical applications, the ability to abstain when uncertain is equally vital for trustworthy depl…

Question Answering