Examining Gender and Racial Bias in Large Vision-Language Models Using a Novel Dataset of Parallel Images

2024/02/08 by Kathleen Fraser, Fraser, Kathleen C., Svetlana Kiritchenko +1 · 14 citations
Social Sciences · #Computation and Language (cs.CL) #Computer Vision and Pattern Recognition (cs.CV) #Computers and Society (cs.CY) #FOS: Computer and information sciences #Vaccine Coverage and Hesitancy

paper · pdf · doi:10.48550/arxiv.2402.05779

openalex publication_date 2024/02/08 · openalex created_date 2024/02/10 · openalex updated_date 2026/07/28

Abstract

Following on recent advances in large language models (LLMs) and subsequent chat models, a new wave of large vision-language models (LVLMs) has emerged. Such models can incorporate images as input in addition to text, and perform tasks such as visual question answering, image captioning, story generation, etc. Here, we examine potential gender and racial biases in such systems, based on the perceived characteristics of the people in the input images. To accomplish this, we present a new dataset PAIRS (PArallel Images for eveRyday Scenarios). The PAIRS dataset contains sets of AI-generated images of people, such that the images are highly similar in terms of background and visual content, but differ along the dimensions of gender (man, woman) and race (Black, white). By querying the LVLMs with such images, we observe significant differences in the responses according to the perceived gender or race of the person depicted.

Cited by

• DSO: Direct Steering Optimization for Bias Mitigation
• Bias Is a Subspace, Not a Coordinate: A Geometric Rethinking of Post-hoc Debiasing in Vision-Language Models
• Back to the Communities: A Mixed-Methods and Community-Driven Evaluation of Cultural Sensitivity in Text-to-Image Models
• Hearsay: Vision-Language Medical Diagnoses Without an Image
• Vision-Language Models Suppress Female Representations Under Ambiguous Input
• The Face of Persuasion: Analyzing Bias and Generating Culture-Aware Ads
• Bias in the Picture: Benchmarking VLMs with Social-Cue News Images and LLM-as-Judge Assessment
• Bias in Gender Bias Benchmarks: How Spurious Features Distort Evaluation
• On the Reproducibility of "FairCLIP: Harnessing Fairness in Vision-Language Learning''
• Vision-Language Models display a strong gender bias
• A Survey on Video Temporal Grounding with Multimodal Large Language Model
• AI Should Sense Better, Not Just Scale Bigger: Adaptive Sensing as a Paradigm Shift
• Toward Valid Measurement Of (Un)fairness For Generative AI: A Proposal For Systematization Through The Lens Of Fair Equality of Chances
• From Individuals to Interactions: Benchmarking Gender Bias in Multimodal Large Language Models from the Lens of Social Relationship

Related

• Uncovering Bias in Large Vision-Language Models with Counterfactuals
• Uncovering Bias in Large Vision-Language Models at Scale with Counterfactuals
• Fairness in AI Systems: Mitigating gender bias from language-vision models
• A Unified Framework and Dataset for Assessing Societal Bias in Vision-Language Models
• Vision-Language Models Generate More Homogeneous Stories for Phenotypically Black Individuals