Inteligência artificial, segurança jurídica e aplicação no poder judiciário brasileiro

Abstract

In the current global scenario, AI (artificial intelligence) is one of the main topics being studied in the legal field. This research problematizes whether, in Brazilian Courts, there would be a minimum legal certainty for a possible and eventual application of AI in decision-making, based on respect for the determination of uniformity of jurisprudence (art. 926 of the CPC). Such a debate can contribute to obtaining a hermeneutics suitable for solving concrete cases, with respect to the Hermeneutical Criticism of Law and even the Economic Analysis of Law. The hypothesis, from the perspective of civil procedural law, points to the lack of a sufficiently secure database to be applied by AI. This approach, without disregarding other possibilities, can contribute to greater adequate legal certainty. The study carried out uses the phenomenological-hermeneutic method and the literature review research methodology. The need to respect legal certainty is defended, with the aim of complying with the legal determination provided for in articles 489 and 926, both of the CPC. Such need and attention is nothing more than respecting the Law itself, as will be demonstrated throughout the research. Furthermore, it will be presented how AI algorithms have been applied to forensic practice and how they can facilitate and assist in the day-to-day work of the Judiciary. It is concluded, therefore, that there is still no sufficiently secure database (jurisprudence and precedents) for the application of AI in decision-making, precisely as a result of the judicial activism practiced by the Courts. Likewise, from a technical and practical point of view, both the judge's activity and that of robots will be subject to the imprecision of natural language (vagueness and ambiguity), leaving room for cases in which there is room for interpretation. These, precisely, are not yet, in their entirety, captured by AI. The legal text established in a general and abstract way cannot, by definition, account for all practical everyday situations, always requiring the discussion of necessary and sufficient conditions. In view of this, even though it is known that AI is impossible to apply to decision-making, respect and attention to the standardization of jurisprudence and so-called precedents would certainly contribute to this idea.