A Unified Framework and Dataset for Assessing Societal Bias in Vision-Language Models

TL;DR

This paper introduces a comprehensive framework and dataset for evaluating societal biases in vision-language models across multiple inference modes, aiming to guide future development of less biased AI systems.

Contribution

It presents a unified evaluation framework and a synthetic dataset to systematically assess gender, race, and age biases in various VLMs across different modalities.

Findings

Bias varies with input-output modalities.

Models show distinct biases across attributes.

Synthetic dataset enables bias benchmarking.

Abstract

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.