The topic of “artificial intelligence” has never been discussed as much as it is today, and there has been a great deal of excitement due to the increased access to tools such as OpenAI’s ChatGPT, which has been operating using the GPT-4 (Generative Pre-training Transformer 4.0) algorithm (1), the latest version powering this tool (2), since March 2023.
The fact is that the use of artificial intelligence systems has been increasing across various sectors, for a variety of purposes, and this is no different in the Brazilian judiciary. These systems are used to streamline routine activities and, above all, to assist in decision-making, which has led to a notable increase in the speed and cost-effectiveness of the administration of justice.
The study “Technology Applied to Conflict Management in the Judiciary with an Emphasis on Artificial Intelligence,” coordinated by the Getúlio Vargas Foundation’s Center for Innovation, Administration, and Judicial Research (CIAPJ/FGV), revealed that, in 2020, more than half of Brazilian courts were already using artificial intelligence systems (3). However, caution is needed, and the guidelines established by the National Council of Justice (CNJ) must be followed to ensure that the use of these systems does not violate constitutionally established rights and guarantees.
The volume of cases filed in the Brazilian justice system, combined with budgetary constraints on the creation of new positions for judges and court staff, presents the Brazilian judiciary with the constant challenge of improving its case management. In this regard, the adoption of algorithmic tools for automating non-decision-making tasks and supporting decision-making has proven essential for achieving greater productivity and speed in the administration of justice, at a lower cost.
The development of these tools has provided a solution for improving the Brazilian judiciary’s case management. The “Victor” system, developed in partnership with the University of Brasília (UnB), is one example. It assists analysts at the Federal Supreme Court (STF) in interpreting appeals and categorizing them by issues of general repercussion, thereby reducing analysis time and conserving human resources. The STF has also implemented the “Victor Robot,” which uses heuristically processed data sets to answer questions from users of the court’s website.
The Superior Court of Justice has also stood out for implementing artificial intelligence-based decision support systems, as described by Mário Augusto Figueiredo de Lacerda Guerreiro: “The Superior Court of Justice (STJ) developed the Socrates 2.0 system, designed to identify and provide relevant information contained in special appeals, such as the constitutional grounds on which the appeal was filed, the legal provisions deemed to have been violated, the case law precedents on the matter, and the legal controversies presented (...)” (4). Other systems developed by the STJ include: “e-Juris, which uses neural network algorithms to identify the legal provisions that were the subject of review and the cited precedents; Logos, designed to identify recurring themes; and Accordes, aimed at selecting similar judgments for the Jurisprudence in Theses product; (...) In partnership with the National School for the Training and Continuing Education of Judges (ENFAM), the STJ developed the Corpus 927 project, a case law search system that compiles the binding decisions, rulings, and guidelines referred to in Article 927 of Law 13.105/2015” (5).
And these are just a few examples of systems adopted by the higher courts. In other areas, there are numerous valuable initiatives by the judiciary that cannot be adequately covered in these few lines. Thus, even though the prospect of incorporating highly sophisticated systems into the administration of justice—such as the aforementioned GPT-4—is on the horizon, there are also challenges to consider, such as ensuring transparency, ethics, and governance in the use of algorithms and machine learning processes.
Obviously, if an error occurs—whether due to poor data collection, faulty processing, or the inability of an algorithmic solution to handle a particular problem—there is a risk that the decision reached will be tainted by bias. In simple terms, algorithmic bias (6) refers to a flaw “in the consequence,” which can cause harm. Nevertheless, it is necessary to consider the “antecedent,” that is, to investigate the causal pathway of the heuristic process so that it is possible to assess whether the flawed decision was made as a result of a failure that occurred in a previous stage, which ultimately tainted the subsequent processing stages. At the regulatory level, however, there remains a lack of legislation worldwide.
In the United States of America, the “Fundamentally Understanding the Usability and Realistic Evolution of Artificial Intelligence Act,” or simply the “Future of AI Act” (7), was introduced on December 12, 2017, which places significant emphasis on the correlation between the concept of AI and the functioning of the human brain, reflecting a conceptual proximity to the idea of “technological singularity.” This document also sets forth ethical guidelines to promote algorithmic development but does not directly address civil liability.
Some more recent documents, such as the 2021 European Artificial Intelligence Act (8) (2021 EU AIA) and the very recent 2022 U.S. Algorithmic Accountability Act (9) (2022 US AAA)—which updated the previous 2019 version (10)—avoid terminological debates regarding the semantic scope of the term “intelligence,” preferring instead to refer to “automated decision-making systems” (11) (Automated Decision Systems, or ADSs) to make explicit the need to establish a civil liability regime applicable to harmful events caused by such systems, and even to reaffirm the importance of establishing ethical parameters for algorithm development.
According to authoritative legal doctrine (12), the cited documents possess qualities that can serve as mutual inspiration, highlighting the importance of proper semantic assimilation (among other issues) for the advancement of discussions even at the global level.
In Brazil, Bills Nos. 5.051/2019, 21/2020, and 872/2021 aim to regulate the issue in general terms (and not merely within the context of the judiciary), prioritizing the establishment of a liability system based on the anachronistic theory of fault, which simply does not make sense for governing such a complex matter. Resolution No. 332/2020 of the National Council of Justice (CNJ), inspired by fundamental principles established by the European Commission for the Efficiency of Justice (CEPEJ) and the Organization for Economic Cooperation and Development (OECD), seeks to ensure that the use of artificial intelligence in the Brazilian judiciary respects fundamental rights and guarantees, without compromising procedural efficiency and cost-effectiveness.
However, in February 2022, the Federal Senate established the commendable “Commission of Jurists Responsible for Assisting in the Drafting of a Substitute Bill on AI” (CJSUBIA), which held several meetings and public hearings; the work on drafting the substitute bill was completed in December 2022 (13).
Undoubtedly, artificial intelligence has great potential to revolutionize the administration of justice in Brazil, making it faster and more efficient. However, it is necessary to ensure that it does not compromise the quality of judicial decisions or violate citizens’ fundamental rights and guarantees. GPT-4 has great potential for future use in the Brazilian judiciary, thanks to its ability to process and analyze large volumes of data efficiently and accurately. With this technology, it will be possible to make judicial proceedings more efficient and faster, since GPT-4 can analyze and interpret texts in an “almost human” way, identifying relevant information in complex documents—such as contracts and petitions—with the added capability of analyzing images, not just text.
In addition, GPT-4 can assist in judicial decision-making, allowing judges and lawyers to quickly access information that is relevant and important to the case at hand. With the use of this technology, it will be possible to conduct more precise and detailed research, making the decision-making process more informed and fair. Finally, GPT-4 also has the potential to be used in citizen service initiatives by the Brazilian judiciary, through smarter and more effective chatbots that can answer questions quickly and accurately, making access to justice easier and more accessible for everyone.
What is undeniable is that such systems, however exciting and sophisticated they may be, are unable to replace the role of judges. Anything that is intrinsic to human nature—to the point of requiring careful analysis of circumstances that go beyond heuristic processing—is certainly not covered by the way these systems operate. Their value, in fact, lies in the efficiency gains they can provide, improving the administration of justice as support systems—and nothing more than that.
(1) OPENAI. Generative Pre-training Transformer 4.0 – GPT-4. Disponível em: https://openai.com/research/gpt-4 Acesso em: 17 mar. 2023.
(2) For a conceptual and evolutionary overview of GPT, see the paper I wrote in 2021, when the first rumors about the potential of GPT-3—the previous version of the algorithm—began to surface, in: FALEIROS JR., José Luiz de Moura. Breves reflexões sobre os impactos jurídicos do algoritmo GPT-3. In: BARBOSA, Mafalda Miranda; BRAGA NETTO, Felipe, SILVA, Michael César; FALEIROS JR., José Luiz de Moura (coord.). Direito digital e inteligência artificial: diálogos entre Brasil e Europa. Indaiatuba, SP: Foco, 2021, p. 521-532.
(3) FUNDAÇÃO GETÚLIO VARGAS. Centro de Inovação, Administração e Pesquisa do Judiciário. Tecnologia Aplicada à Gestão de Conflitos no Poder Judiciário com Ênfase em Inteligência Artificial, 2020. Disponível em: https://portal.fgv.br/eventos/webinar-i-inteligencia-artificial-aplicada-gestao-conflitos-ambito-poder-judiciario-1o-forum. Acesso em: 17 mar. 2023.
(4) DANKS, David; LONDON, Alex John. Algorithmic bias in autonomous systems. Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence (IJCAI-17), Viena, p. 4691-4697, 2017. Disponível em: https://www.ijcai.org/Proceedings/2017/ Acesso em: 17 mar. 2023.
(5) GUERREIRO, Mário Augusto Figueiredo de Lacerda. Inovações na adoção da inteligência artificial pelo Poder Judiciário brasileiro. In: BARBOSA, Mafalda Miranda; BRAGA NETTO, Felipe, SILVA, Michael César; FALEIROS JR., José Luiz de Moura (coord.). Direito digital e inteligência artificial: diálogos entre Brasil e Europa. Indaiatuba, SP: Foco, 2021, p. 511-512.
(6) Ibid.
(7) ESTADOS UNIDOS DA AMÉRICA. House of Representatives. House Resolution No. 4625, Dec. 12, 2017. FUTURE of Artificial Intelligence Act. Disponível em: https://www.congress.gov/115/bills/hr4625/BILLS-115hr4625ih.pdf Acesso em: 17 mar. 2023.
(8) EUROPA. European Commission. Artificial Intelligence Act. 2021/0106(COD), abr. 2021. Disponível em: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A52021PC0206 Acesso em: 17 mar. 2023.
(9) ESTADOS UNIDOS DA AMÉRICA. House of Representatives. House Resolution No. 6580, Feb. 3, 2022. Algorithmic Accountability Act of 2022. Disponível em: https://www.congress.gov/bill/117th-congress/house-bill/6580/text Acesso em: 17 mar. 2023.
(10) ESTADOS UNIDOS DA AMÉRICA. House of Representatives. House Resolution No. 2231, Apr. 10, 2019. Algorithmic Accountability Act of 2019. Disponível em: https://www.congress.gov/116/bills/hr2231/BILLS-116hr2231ih.pdf Acesso em: 17 mar. 2023.
(11) See SELBST, Andrew. An institutional view of algorithmic impact assessments. Harvard Journal of Law & Technology, Cambridge, v. 35, 2021. Disponível em: https://ssrn.com/abstract=3867634 Acesso em: 17 mar. 2023.
(12) MÖKANDER, Jakob; JUNEJA, Prathm; WATSON, David S.; FLORIDI, Luciano. The US Algorithmic Accountability Act of 2022 vs. The EU Artificial Intelligence Act: what can they learn from each other? Minds and Machines, Cham: Springer, v. 22, p. 1-9, jun. 2022. Disponível em: https://doi.org/10.1007/s11023-022-09612-y Acesso em: 17 mar. 2023.
(13) BRASIL. Senado Federal. Atividade Legislativa. Comissão de Juristas responsável por subsidiar elaboração de substitutivo sobre IA (CJSUBIA). Disponível em: https://legis.senado.leg.br/comissoes/comissao?codcol=2504 Acesso em: 17 mar. 2023.
Article originally written on March 17, 2023