Política criminal atuarial e a tomada de decisão judicial por sistemas de IA: a análise da simetria de sua relação como medida para sua refutação

Abstract

This work was developed based on the analysis of Actuarial Criminal Policy. It is described by Maurício Dieter (2023) as being a managerialist way of dealing with crime, which categorizes social groups according to the risk they represent, in order to calibrate repression measures and prevent crime through selective neutralization. Our objective was to explore the relationship maintained between Actuarial Criminal Policy, the implementation of the use of artificial intelligence (AI) in the brazilian Judiciary and the threat to guarantee principles, as containment filters for state punitive power. To this end, we draw on the constructions and criticisms made by the author of the criminal policy model in question, which appears to violate constitutional principles such as culpability, legality, equality, proportionality, individualization and humanization of punishment. Furthermore, we add to the thematic equation the lens of AI systems as possible mechanisms for instrumentalizing political-criminological actuarial logic today, especially when used as decision-making mechanisms. Along these lines, the question to be answered is: “With the use of AI systems by the Judiciary, are there legal guarantee mechanisms to curb the intrinsic selectivity of the punitive power related to the Actuarial Criminal Policy, in order to ensure the fundamental rights of those criminally prosecuted?” The methodology adopted is based on predominantly bibliographical research in Dieter (2023), Toledo and Pessoa (2023) and Bitencourt (2022). Next, in empirical research at the National Council of Justice (CNJ), we carried out a brief analysis of AI systems used by the brazilian Judiciary. In empirical research of a documentary nature, in a normative administrative act issued by the Executive Branch, we evaluated the National Criminal and Penitentiary Policy Plan (PNPCP) 2020-2023 (Brasil, 2019), in order to identify possible institutes governed by actuarial logic. Regarding the results found, it is clear that the use of AI by the brazilian Judiciary is a reality under construction and the adoption of tools guided by actuarialism in the criminal sector can already be found in the PNPCP 2020-2023. In conclusion, we offer a proposal that seeks to protect the Democratic Rule of Law, in the face of the implementation of economic-actuarial logic, which, in contemporary capitalism, naturalizes criminal selectivity and, through the diagnosis of risk, legitimizes repressive action based on mere calculations of probability.