BIGbench: A Unified Benchmark for Evaluating Multi-dimensional Social Biases in Text-to-Image Models

TL;DR

BIGbench is a comprehensive benchmark that classifies and evaluates social biases in text-to-image models across multiple dimensions, using advanced models for automated assessment, enabling detailed bias analysis and comparison.

Contribution

Introduces BIGbench, a novel unified benchmark with a multi-dimensional bias classification and automated evaluation for T2I models, advancing bias analysis methodology.

Findings

BIGbench effectively identifies various biases in T2I models.

Automated evaluations align well with human judgments.

Analysis reveals new bias research directions.

Abstract

Text-to-Image (T2I) generative models are becoming increasingly crucial due to their ability to generate high-quality images, but also raise concerns about social biases, particularly in human image generation. Sociological research has established systematic classifications of bias. Yet, existing studies on bias in T2I models largely conflate different types of bias, impeding methodological progress. In this paper, we introduce BIGbench, a unified benchmark for Biases of Image Generation, featuring a carefully designed dataset. Unlike existing benchmarks, BIGbench classifies and evaluates biases across four dimensions to enable a more granular evaluation and deeper analysis. Furthermore, BIGbench applies advanced multi-modal large language models to achieve fully automated and highly accurate evaluations. We apply BIGbench to evaluate eight representative T2I models and three debiasing methods. Our human evaluation results by trained evaluators from different races underscore BIGbench's effectiveness in aligning images and identifying various biases. Moreover, our study also reveals new research directions about biases with insightful analysis of our results. Our work is openly accessible at https://github.com/BIGbench2024/BIGbench2024/.