USE OF ARTIFICIAL INTELLIGENCE (AI) IN THE JUDICIAL PROCESS: ANALYSIS OF THE ETHICAL AND PROCEDURAL IMPACTS OF AI USE IN JUDICIAL DECISIONS IN LIGHT OF THE BRAZILIAN LEGAL SYSTEM.

Abstract

This article investigates the growing implementation of Artificial Intelligence (AI) and automation systems in the Brazilian Judiciary, focusing on the impacts of this technology on judicial decision-making. Based on bibliographical, documentary, and empirical research, the current panorama of the courts is analyzed, revealing the veiled delegation of decision-making functions to machines under the justification of procedural efficiency and speed. The study addresses two central problems: the extreme opacity of the algorithms operated by the courts (the "black box") and the exponential reproduction of human cognitive biases, disguised as algorithmic biases. It demonstrates the structural incompatibility between the mathematical logic of AI data-driven standardization and the legal requirement to individualize specific cases, in addition to the inherent inability of machines to formulate autonomous ethical-legal value judgments. The research notes that replacing human judgment with statistical inferences not only threatens due process, the adversarial system, and the right to a broad defense, but also risks institutionalizing historical discrimination. As alternatives, the text lists proposals to mitigate these damages, such as the requirement for external audits, hybrid decision architectures, algorithmic transparency, and the right to human review, in accordance with Bill No. 2,338/2023. The article concludes that, due to the severe risks to fundamental rights and the Democratic Rule of Law, the current use of AI in the Judiciary should be restricted to the automation of bureaucratic routines, making its application in the merits of judicial decisions reckless without proper democratic control.