Image Representations Learned With Unsupervised Pre-Training Contain Human-like Biases

TL;DR

This paper investigates whether unsupervised computer vision models trained on web images automatically learn social biases, revealing that they do encode racial, gender, and intersectional biases similar to human stereotypes.

Contribution

The study introduces a novel method to quantify social biases in unsupervised image representations and demonstrates that these models learn biases aligned with human stereotypes from web data.

Findings

Unsupervised models learn racial, gender, and intersectional biases.

Models replicate 8 social biases documented in psychology.

Biases in models reflect stereotypes present in online image datasets.

Abstract

Recent advances in machine learning leverage massive datasets of unlabeled images from the web to learn general-purpose image representations for tasks from image classification to face recognition. But do unsupervised computer vision models automatically learn implicit patterns and embed social biases that could have harmful downstream effects? We develop a novel method for quantifying biased associations between representations of social concepts and attributes in images. We find that state-of-the-art unsupervised models trained on ImageNet, a popular benchmark image dataset curated from internet images, automatically learn racial, gender, and intersectional biases. We replicate 8 previously documented human biases from social psychology, from the innocuous, as with insects and flowers, to the potentially harmful, as with race and gender. Our results closely match three hypotheses about intersectional bias from social psychology. For the first time in unsupervised computer vision, we also quantify implicit human biases about weight, disabilities, and several ethnicities. When compared with statistical patterns in online image datasets, our findings suggest that machine learning models can automatically learn bias from the way people are stereotypically portrayed on the web.