Saved in:
Bibliographic Details
Main Authors: Brand, Timo, Förster, Henry, Kobourov, Stephen G., Miller, Jacob
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2511.03617
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918187372642304
author Brand, Timo
Förster, Henry
Kobourov, Stephen G.
Miller, Jacob
author_facet Brand, Timo
Förster, Henry
Kobourov, Stephen G.
Miller, Jacob
contents We evaluate how visualizations can influence the judgment of MLLMs about the presence or absence of bridges in a network. We show that the inclusion of visualization improves confidence over a structured text-based input that could theoretically be helpful for answering the question. On the other hand, we observe that standard visualization techniques create a strong bias towards accepting or refuting the presence of a bridge -- independently of whether or not a bridge actually exists in the network. While our results indicate that the inclusion of visualization techniques can effectively influence the MLLM's judgment without compromising its self-reported confidence, they also imply that practitioners must be careful of allowing users to include visualizations in generative AI applications so as to avoid undesired hallucinations.
format Preprint
id arxiv_https___arxiv_org_abs_2511_03617
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Visualization Biases MLLM's Decision Making in Network Data Tasks
Brand, Timo
Förster, Henry
Kobourov, Stephen G.
Miller, Jacob
Graphics
Artificial Intelligence
We evaluate how visualizations can influence the judgment of MLLMs about the presence or absence of bridges in a network. We show that the inclusion of visualization improves confidence over a structured text-based input that could theoretically be helpful for answering the question. On the other hand, we observe that standard visualization techniques create a strong bias towards accepting or refuting the presence of a bridge -- independently of whether or not a bridge actually exists in the network. While our results indicate that the inclusion of visualization techniques can effectively influence the MLLM's judgment without compromising its self-reported confidence, they also imply that practitioners must be careful of allowing users to include visualizations in generative AI applications so as to avoid undesired hallucinations.
title Visualization Biases MLLM's Decision Making in Network Data Tasks
topic Graphics
Artificial Intelligence
url https://arxiv.org/abs/2511.03617