@inproceedings{ji-etal-2025-calibrating,
title = "Calibrating Verbal Uncertainty as a Linear Feature to Reduce Hallucinations",
author = "Ji, Ziwei and
Yu, Lei and
Koishekenov, Yeskendir and
Bang, Yejin and
Hartshorn, Anthony and
Schelten, Alan and
Zhang, Cheng and
Fung, Pascale and
Cancedda, Nicola",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.emnlp-main.187/",
pages = "3769--3793",
ISBN = "979-8-89176-332-6",
abstract = "LLMs often adopt an assertive language style also when making false claims. Such ``overconfident hallucinations'' mislead users and erode trust. Achieving the ability to express in language the actual degree of uncertainty around a claim is therefore of great importance. We find that ``verbal uncertainty'' is governed by a single linear feature in the representation space of LLMs, and shows that this has only moderate correlation with the actual ``semantic uncertainty'' of the model. We apply this insight and show that (1) the mismatch between semantic and verbal uncertainty is a better predictor of hallucinations than semantic uncertainty alone and (2) we can intervene on verbal uncertainty at inference time and reduce confident hallucinations on short-form answers, achieving an average relative reduction of {\textasciitilde}30{\%}."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="ji-etal-2025-calibrating">
<titleInfo>
<title>Calibrating Verbal Uncertainty as a Linear Feature to Reduce Hallucinations</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ziwei</namePart>
<namePart type="family">Ji</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lei</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yeskendir</namePart>
<namePart type="family">Koishekenov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yejin</namePart>
<namePart type="family">Bang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Anthony</namePart>
<namePart type="family">Hartshorn</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alan</namePart>
<namePart type="family">Schelten</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Cheng</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pascale</namePart>
<namePart type="family">Fung</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nicola</namePart>
<namePart type="family">Cancedda</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Christos</namePart>
<namePart type="family">Christodoulopoulos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Carolyn</namePart>
<namePart type="family">Rose</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Violet</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Suzhou, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-332-6</identifier>
</relatedItem>
<abstract>LLMs often adopt an assertive language style also when making false claims. Such “overconfident hallucinations” mislead users and erode trust. Achieving the ability to express in language the actual degree of uncertainty around a claim is therefore of great importance. We find that “verbal uncertainty” is governed by a single linear feature in the representation space of LLMs, and shows that this has only moderate correlation with the actual “semantic uncertainty” of the model. We apply this insight and show that (1) the mismatch between semantic and verbal uncertainty is a better predictor of hallucinations than semantic uncertainty alone and (2) we can intervene on verbal uncertainty at inference time and reduce confident hallucinations on short-form answers, achieving an average relative reduction of ~30%.</abstract>
<identifier type="citekey">ji-etal-2025-calibrating</identifier>
<location>
<url>https://aclanthology.org/2025.emnlp-main.187/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>3769</start>
<end>3793</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Calibrating Verbal Uncertainty as a Linear Feature to Reduce Hallucinations
%A Ji, Ziwei
%A Yu, Lei
%A Koishekenov, Yeskendir
%A Bang, Yejin
%A Hartshorn, Anthony
%A Schelten, Alan
%A Zhang, Cheng
%A Fung, Pascale
%A Cancedda, Nicola
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-332-6
%F ji-etal-2025-calibrating
%X LLMs often adopt an assertive language style also when making false claims. Such “overconfident hallucinations” mislead users and erode trust. Achieving the ability to express in language the actual degree of uncertainty around a claim is therefore of great importance. We find that “verbal uncertainty” is governed by a single linear feature in the representation space of LLMs, and shows that this has only moderate correlation with the actual “semantic uncertainty” of the model. We apply this insight and show that (1) the mismatch between semantic and verbal uncertainty is a better predictor of hallucinations than semantic uncertainty alone and (2) we can intervene on verbal uncertainty at inference time and reduce confident hallucinations on short-form answers, achieving an average relative reduction of ~30%.
%U https://aclanthology.org/2025.emnlp-main.187/
%P 3769-3793
Markdown (Informal)
[Calibrating Verbal Uncertainty as a Linear Feature to Reduce Hallucinations](https://aclanthology.org/2025.emnlp-main.187/) (Ji et al., EMNLP 2025)
ACL
- Ziwei Ji, Lei Yu, Yeskendir Koishekenov, Yejin Bang, Anthony Hartshorn, Alan Schelten, Cheng Zhang, Pascale Fung, and Nicola Cancedda. 2025. Calibrating Verbal Uncertainty as a Linear Feature to Reduce Hallucinations. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 3769–3793, Suzhou, China. Association for Computational Linguistics.