A utilização da inteligência artificial no judiciário: Possibilidade de discriminação no sistema penal?

Abstract

This study analyzes the application of artificial intelligence (AI) in the Brazilian judicial system, with an emphasis on the risks of algorithmic discrimination in the criminal justice system. Based on the premise that AI tools are not neutral, as their algorithms, fed by historical data, reproduce and reinforce preexisting social biases, the research investigates how this technology can reinforce inequalities in the criminal justice system. The question this research proposed is: can the use of artificial intelligence systems as a tool generate algorithmic discrimination, and how might this occur? The general objective was to study the use of artificial intelligence in the Brazilian Judicial System and the possibility of algorithmic discrimination in the criminal justice system. Furthermore, the specific objectives sought to: a) understand what artificial intelligence is and its compositions; b) the state of the Brazilian judicial system in light of new technologies; c) the possibility of algorithmic discrimination. The methodological approach combined Karl Popper's hypothetical-deductive method, hypothesis-based literature review, and empirical data analysis, along with document research from sources such as CNJ reports, academic articles, and national and international case investigations. The findings showed that, although AI offers advantages such as faster processes and lower costs, it also reproduces patterns of discrimination. In Brazil, the use of facial recognition by the police revealed high error rates for Black people, transgender people, and people from peripheral areas, leading to unfair arrests. Furthermore, the lack of transparency in the algorithms and the lack of adequate regulation exacerbated these problems, making oversight and judicial control more difficult. The conclusions supported the initial hypothesis: AI, when used without ethical guidelines and legal safeguards, tends to maintain and intensify structural discrimination, especially in a penal system already characterized by selectivity and racism. Therefore, the innovations proposed by AI require regulations prepared for the impacts arising from their use, applying ethical criteria that prevent the reinstatement of biases present in previous decisions, that is, preventing automation from deepening social injustices rather than resolving them. Regulation of AI in the judicial system should focus on accountability, social participation, and the correction of biases, ensuring that technological advances represent social evolution and do not compromise justice and equity.