Inteligência artificial e ética: uma análise kantiana e utilitarista da centralidade humana nas decisões judiciais

Abstract

This monograph analyzes the ethical challenges of building and using artificial intelligence within the Brazilian Judiciary, with a central focus on the question of human centrality. To this end, two ethical perspectives were selected – Kantian deontological ethics and John Stuart Mill's consequentialist-utilitarian approach. The research addresses the evolution of AI, from its origins to the present day, discusses the question of artificial consciousness, and examines ethical challenges of artificial intelligence, with emphasis on algorithmic bias, responsibility, transparency, and hallucination. The research also reflects on the possibility of delegating judicial decisions to machines, concluding that this hypothesis should not be pursued given various ethical and practical issues it addresses. Based on this conclusion, it proposes the recognition of a fundamental right to judgment by humans, as an extension of the principle of human dignity. Through comparative analysis of Brazilian and European AI legislation, the work highlights human centrality as a guiding principle of normative texts. As a practical contribution, it presents a hybrid ethical risk framework complementary to Resolution 615 of the National Council of Justice, structured in four increasing levels of AI autonomy, always preserving effective human supervision. The work concludes by arguing that artificial intelligence should function as a tool for empowering the human judge, never as a substitute.