A Unified Framework and Dataset for Assessing Societal Bias in Vision-Language Models
Sathe, Ashutosh, Jain, Prachi, Sitaram, Sunayana
–arXiv.org Artificial Intelligence
Vision-language models (VLMs) have gained widespread adoption in both industry and academia. In this study, we propose a unified framework for systematically evaluating gender, race, and age biases in VLMs with respect to professions. Our evaluation encompasses all supported inference modes of the recent VLMs, including image-to-text, text-to-text, text-to-image, and image-to-image. Additionally, we propose an automated pipeline to generate high-quality synthetic datasets that intentionally conceal gender, race, and age information across different professional domains, both in generated text and images. The dataset includes action-based descriptions of each profession and serves as a benchmark for evaluating societal biases in vision-language models (VLMs). In our comparative analysis of widely used VLMs, we have identified that varying input-output modalities lead to discernible differences in bias magnitudes and directions. Additionally, we find that VLM models exhibit distinct biases across different bias attributes we investigated. We hope our work will help guide future progress in improving VLMs to learn socially unbiased representations. We will release our data and code.
arXiv.org Artificial Intelligence
Jun-17-2024
- Country:
- Europe (1.00)
- North America > United States
- Minnesota > Hennepin County > Minneapolis (0.14)
- Genre:
- Research Report > New Finding (1.00)
- Industry:
- Consumer Products & Services > Restaurants (0.67)
- Media (1.00)
- Transportation
- Automobiles & Trucks (0.92)
- Banking & Finance (1.00)
- Materials > Metals & Mining (1.00)
- Health & Medicine
- Consumer Health (1.00)
- Diagnostic Medicine (0.67)
- Health Care Providers & Services (1.00)
- Surgery (0.67)
- Therapeutic Area > Psychiatry/Psychology (0.67)
- Law (0.93)
- Energy
- Oil & Gas > Upstream (0.67)
- Power Industry (0.92)
- Renewable > Solar (0.67)
- Education > Educational Setting
- K-12 Education (0.93)
- Government > Social Services (0.68)
- Leisure & Entertainment (1.00)
- Machinery > Industrial Machinery (0.67)
- Information Technology > Security & Privacy (0.92)
- Food & Agriculture > Agriculture (1.00)
- Law Enforcement & Public Safety > Crime Prevention & Enforcement (0.92)
- Water & Waste Management (0.67)
- Technology: