Automating Justice: AI and the Courts

Abstract

This case study explores the development and deployment of an AI-driven adjudication system by Augmented Justice Inc. (AJI) to address the overwhelming backlog of U.S. immigration cases, particularly those involving asylum seekers and humanitarian parole. Faced with over 1.8 million pending cases and years-long delays, AJI developed a supervised decision-tree model to evaluate applicant data, recommend case outcomes, and generate detailed, auditable decision rationales. The system aimed to enhance efficiency, transparency, and fairness, particularly by identifying bias in prior rulings. After extensive data training, legal auditing, and interface design, AJI's software was adopted by the Department of Homeland Security. It successfully reduced hearing delays and court time while also providing applicants with predictive outcome scores and supporting documentation. However, the case raises pressing concerns about dehumanization, accountability, bias reproduction, and the limits of automation in justice. While the AI system improved efficiency and flagged patterns of systemic discrimination, it also prompted ethical questions about deterrence, due process, and the legitimacy of replacing human judgment with algorithmic logic in humanitarian legal decisions.