@inproceedings{borszukovszki-etal-2025-know,
title = "Know What You do Not Know: Verbalized Uncertainty Estimation Robustness on Corrupted Images in Vision-Language Models",
author = "Borszukovszki, Mirko and
De Jong, Ivo Pascal and
Valdenegro-Toro, Matias",
editor = "Cao, Trista and
Das, Anubrata and
Kumarage, Tharindu and
Wan, Yixin and
Krishna, Satyapriya and
Mehrabi, Ninareh and
Dhamala, Jwala and
Ramakrishna, Anil and
Galystan, Aram and
Kumar, Anoop and
Gupta, Rahul and
Chang, Kai-Wei",
booktitle = "Proceedings of the 5th Workshop on Trustworthy NLP (TrustNLP 2025)",
month = may,
year = "2025",
address = "Albuquerque, New Mexico",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.trustnlp-main.16/",
doi = "10.18653/v1/2025.trustnlp-main.16",
pages = "247--265",
ISBN = "979-8-89176-233-6",
abstract = "To leverage the full potential of Large Language Models (LLMs) it is crucial to have some information on their answers' uncertainty. This means that the model has to be able to quantify how certain it is in the correctness of a given response. Bad uncertainty estimates can lead to overconfident wrong answers undermining trust in these models. Quite a lot of research has been done on language models that work with text inputs and provide text outputs. Still, since the visual capabilities have been added to these models recently, there has not been much progress on the uncertainty of Visual Language Models (VLMs). We tested three state-of-the-art VLMs on corrupted image data. We found that the severity of the corruption negatively impacted the models' ability to estimate their uncertainty and the models also showed overconfidence in most of the experiments."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="borszukovszki-etal-2025-know">
<titleInfo>
<title>Know What You do Not Know: Verbalized Uncertainty Estimation Robustness on Corrupted Images in Vision-Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Mirko</namePart>
<namePart type="family">Borszukovszki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ivo</namePart>
<namePart type="given">Pascal</namePart>
<namePart type="family">De Jong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Matias</namePart>
<namePart type="family">Valdenegro-Toro</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-05</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 5th Workshop on Trustworthy NLP (TrustNLP 2025)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Trista</namePart>
<namePart type="family">Cao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Anubrata</namePart>
<namePart type="family">Das</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tharindu</namePart>
<namePart type="family">Kumarage</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yixin</namePart>
<namePart type="family">Wan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Satyapriya</namePart>
<namePart type="family">Krishna</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ninareh</namePart>
<namePart type="family">Mehrabi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jwala</namePart>
<namePart type="family">Dhamala</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Anil</namePart>
<namePart type="family">Ramakrishna</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Aram</namePart>
<namePart type="family">Galystan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Anoop</namePart>
<namePart type="family">Kumar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rahul</namePart>
<namePart type="family">Gupta</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kai-Wei</namePart>
<namePart type="family">Chang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Albuquerque, New Mexico</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-233-6</identifier>
</relatedItem>
<abstract>To leverage the full potential of Large Language Models (LLMs) it is crucial to have some information on their answers’ uncertainty. This means that the model has to be able to quantify how certain it is in the correctness of a given response. Bad uncertainty estimates can lead to overconfident wrong answers undermining trust in these models. Quite a lot of research has been done on language models that work with text inputs and provide text outputs. Still, since the visual capabilities have been added to these models recently, there has not been much progress on the uncertainty of Visual Language Models (VLMs). We tested three state-of-the-art VLMs on corrupted image data. We found that the severity of the corruption negatively impacted the models’ ability to estimate their uncertainty and the models also showed overconfidence in most of the experiments.</abstract>
<identifier type="citekey">borszukovszki-etal-2025-know</identifier>
<identifier type="doi">10.18653/v1/2025.trustnlp-main.16</identifier>
<location>
<url>https://aclanthology.org/2025.trustnlp-main.16/</url>
</location>
<part>
<date>2025-05</date>
<extent unit="page">
<start>247</start>
<end>265</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Know What You do Not Know: Verbalized Uncertainty Estimation Robustness on Corrupted Images in Vision-Language Models
%A Borszukovszki, Mirko
%A De Jong, Ivo Pascal
%A Valdenegro-Toro, Matias
%Y Cao, Trista
%Y Das, Anubrata
%Y Kumarage, Tharindu
%Y Wan, Yixin
%Y Krishna, Satyapriya
%Y Mehrabi, Ninareh
%Y Dhamala, Jwala
%Y Ramakrishna, Anil
%Y Galystan, Aram
%Y Kumar, Anoop
%Y Gupta, Rahul
%Y Chang, Kai-Wei
%S Proceedings of the 5th Workshop on Trustworthy NLP (TrustNLP 2025)
%D 2025
%8 May
%I Association for Computational Linguistics
%C Albuquerque, New Mexico
%@ 979-8-89176-233-6
%F borszukovszki-etal-2025-know
%X To leverage the full potential of Large Language Models (LLMs) it is crucial to have some information on their answers’ uncertainty. This means that the model has to be able to quantify how certain it is in the correctness of a given response. Bad uncertainty estimates can lead to overconfident wrong answers undermining trust in these models. Quite a lot of research has been done on language models that work with text inputs and provide text outputs. Still, since the visual capabilities have been added to these models recently, there has not been much progress on the uncertainty of Visual Language Models (VLMs). We tested three state-of-the-art VLMs on corrupted image data. We found that the severity of the corruption negatively impacted the models’ ability to estimate their uncertainty and the models also showed overconfidence in most of the experiments.
%R 10.18653/v1/2025.trustnlp-main.16
%U https://aclanthology.org/2025.trustnlp-main.16/
%U https://doi.org/10.18653/v1/2025.trustnlp-main.16
%P 247-265
Markdown (Informal)
[Know What You do Not Know: Verbalized Uncertainty Estimation Robustness on Corrupted Images in Vision-Language Models](https://aclanthology.org/2025.trustnlp-main.16/) (Borszukovszki et al., TrustNLP 2025)
ACL