Judging by Appearances? Auditing and Intervening Vision-Language Models for Bail Prediction
This work addresses the risk of biased bail decisions in legal systems using VLMs, offering incremental improvements through targeted interventions.
The study audited vision-language models (VLMs) for bail prediction and found they performed poorly across intersectional groups, often denying bail to deserving individuals with high confidence. By incorporating legal precedents via RAG and fine-tuning, the interventions substantially improved bail prediction performance.
Large language models (LLMs) have been extensively used for legal judgment prediction tasks based on case reports and crime history. However, with a surge in the availability of large vision language models (VLMs), legal judgment prediction systems can now be made to leverage the images of the criminals in addition to the textual case reports/crime history. Applications built in this way could lead to inadvertent consequences and be used with malicious intent. In this work, we run an audit to investigate the efficiency of standalone VLMs in the bail decision prediction task. We observe that the performance is poor across multiple intersectional groups and models \textit{wrongly deny bail to deserving individuals with very high confidence}. We design different intervention algorithms by first including legal precedents through a RAG pipeline and then fine-tuning the VLMs using innovative schemes. We demonstrate that these interventions substantially improve the performance of bail prediction. Our work paves the way for the design of smarter interventions on VLMs in the future, before they can be deployed for real-world legal judgment prediction.