Examining Gender and Racial Bias in Large Vision-Language Models Using a Novel Dataset of Parallel Images
This work addresses bias in AI systems, which is a critical issue for fairness and equity, but it is incremental as it applies existing bias analysis methods to a new dataset.
The study investigated gender and racial biases in large vision-language models by creating the PAIRS dataset of AI-generated parallel images varying in gender and race, and found significant differences in model responses based on these characteristics.
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.