Saved in:
Bibliographic Details
Main Authors: Lim, Jaehyuk, Lee, Bruce W.
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2408.09111
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910650380320768
author Lim, Jaehyuk
Lee, Bruce W.
author_facet Lim, Jaehyuk
Lee, Bruce W.
contents This paper examines a phenomenon in multimodal language models where pre-marked options in question images can significantly influence model responses. Our study employs a systematic methodology to investigate this effect: we present models with images of multiple-choice questions, which they initially answer correctly, then expose the same model to versions with pre-marked options. Our findings reveal a significant shift in the models' responses towards the pre-marked option, even when it contradicts their answers in the neutral settings. Comprehensive evaluations demonstrate that this agreeableness bias is a consistent and quantifiable behavior across various model architectures. These results show potential limitations in the reliability of these models when processing images with pre-marked options, raising important questions about their application in critical decision-making contexts where such visual cues might be present.
format Preprint
id arxiv_https___arxiv_org_abs_2408_09111
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Measuring Agreeableness Bias in Multimodal Models
Lim, Jaehyuk
Lee, Bruce W.
Artificial Intelligence
Computation and Language
Computer Vision and Pattern Recognition
Human-Computer Interaction
This paper examines a phenomenon in multimodal language models where pre-marked options in question images can significantly influence model responses. Our study employs a systematic methodology to investigate this effect: we present models with images of multiple-choice questions, which they initially answer correctly, then expose the same model to versions with pre-marked options. Our findings reveal a significant shift in the models' responses towards the pre-marked option, even when it contradicts their answers in the neutral settings. Comprehensive evaluations demonstrate that this agreeableness bias is a consistent and quantifiable behavior across various model architectures. These results show potential limitations in the reliability of these models when processing images with pre-marked options, raising important questions about their application in critical decision-making contexts where such visual cues might be present.
title Measuring Agreeableness Bias in Multimodal Models
topic Artificial Intelligence
Computation and Language
Computer Vision and Pattern Recognition
Human-Computer Interaction
url https://arxiv.org/abs/2408.09111