Inteligência Artificial e gênero no Poder Judiciário brasileiro: desafios éticos e diretrizes para uma regulação feminista e inclusiva

Abstract

Artificial Intelligence (AI) has consolidated itself as one of the most transformative technologies of the twenty-first century, profoundly reshaping the production of knowledge, social relations, and institutional structures. Within the field of applied social sciences, its relevance is particularly pronounced when considering its incorporation into systems of power and decision-making, such as the Judiciary. The increasing use of algorithms and automated systems for case screening, classification, and judicial decision support has often been justified through the discourse of efficiency and technical rationality. However, contemporary academic studies and recent empirical evidence reveal that such technologies are not neutral; rather, they may reproduce and exacerbate pre-existing inequalities, especially those related to gender, race, and class. In this context, the present doctoral dissertation investigates how artificial intelligence, when applied to the Brazilian Judiciary, may reproduce or intensify gender-based discrimination, and to what extent the law can operate effectively in preventing such biases and reorienting technology toward the promotion of ethics and social justice. The study is grounded in the hypothesis that the law can serve as an effective regulatory instrument, provided that it simultaneously integrates technical safeguards (e.g., fairness, transparency, explainability, and auditability), regulatory safeguards (e.g., data governance, human rights, and constitutional principles of due process), and political safeguards (e.g., awareness, critical education, and gender-sensitive technological literacy). These dimensions are understood as interdependent and essential for ensuring that technology is aligned with principles of social justice and inclusion. The main objective is to propose protocols and guidelines that guarantee the fair, transparent, and inclusive use of AI in the Brazilian Judiciary, based on the identification of gender biases potentially reproduced or amplified by such systems and on the analysis of the law’s regulatory capacity to mitigate them. Methodologically, the research adopts a qualitative and interpretative approach, exploratory and applied in nature, structured as a case study of the Brazilian Judiciary. Three principal methodological strategies were employed: (a) a comprehensive literature review in national and international academic databases; (b) documentary analysis of normative acts, resolutions, and reports issued by the Conselho Nacional de Justiça (CNJ) and judicial institutions; and (c) content analysis, as proposed by Bardin (2016), applied to institutional, jurisprudential, and technical documents. The temporal scope spans from 2018 to 2025, a period corresponding to the consolidation of digital transformation policies and the initial implementation of AI projects within the judicial system. The findings indicate that AI, when integrated into a judiciary still characterized by gender asymmetries and a predominantly male legal culture, tends to replicate structural inequalities. Nonetheless, the research demonstrates that the law can function as a mitigating mechanism, particularly when informed by feminist and intersectional perspectives, promoting the development of technical, regulatory, pedagogical, and political instruments aimed at algorithmic justice. Two ethical protocols are proposed to operationalize the three dimensions of safeguards: (1) a protocol for developers and technical teams, translating technical safeguards into practices of ethical planning, data governance, fairness testing, explainability, and continuous auditing; and (2) a protocol for judges, advisors, and users, implementing regulatory and political safeguards through responsible and transparent AI use, gender-sensitive adjudication, and the preservation of human decision-making autonomy. As both a theoretical and practical contribution, the dissertation proposes a feminist and inclusive model of algorithmic governance for AI within the justice system, structured around interdependent technical, regulatory, and political safeguards and exemplified through the proposed protocols. These instruments constitute the main applied contribution of the study, offering operational and verifiable references for the ethical development, deployment, and oversight of intelligent systems in the Judiciary. Ultimately, this research provides a conceptual and normative framework capable of informing public policies and institutional practices committed to equity, ethics, and social justice in the use of judicial technologies.