The Ethics of Predictive Justice: Can Machine Learning Harmonize Fairnessand Efficiency in Legal Adjudication?

Abstract

Machine-learning (ML) technologies are increasingly being adopted by courts and administrative bodies to improve the speed, consistency, and predictability of adjudication. Proponents claim these predictive-justice systems can advance fairness by reducing human bias, while critics warn they may entrench discrimination through opaque and immutable algorithmic classifications. This article examines whether predictive justice can genuinely harmonize fairness and efficiency within constitutional and human-rights frameworks. Drawing on doctrinal analysis, empirical research, and comparative perspectives from the United States and the European Union, it argues that algorithmic immutability—the persistence of ML-generated classifications—creates new categories of disadvantage beyond the reach of existing law. The study concludes with policy and doctrinal reforms emphasizing accountability, transparency, and contestability to ensure that machine learning enhances, rather than erodes, the legitimacy of legal adjudication.