This article examines the relationship between AI, ethics, and justice, arguing that justice, when grounded in logical constructs and empirical validation, provides a more suitable framework for training and using AI than traditional ethical theories. It explores the evolution of justice from historical philosophical traditions to modern computational models, emphasizing the role of AI in measuring consequences, adapting them to human-set goals, and conducting safe decision-making processes that align with the short and long-term interests of individuals and societies AI’s design must support. While ethics remains fundamentally human-centred, AI can support justice through structured choices and empirical analysis. The paper addresses challenges in integrating AI with justice, critiques alternative ethical frameworks, and provides policy recommendations for optimizing AI-driven justice systems.