Regulação da inteligência artificial no Brasil: análise dos projetos de lei em tramitação na Câmara dos Deputados e no Senado Federal

Abstract

With the increase in computing power and availability of data in electronic media, the use of artificial intelligence systems has been boosted, making their application recurrent in several fields, especially robotics, computer vision, speech recognition and natural language processing. On the other hand, with the advances resulting from the introduction of artificial intelligence systems in society, questions related to the risks and negative impacts that can derive from this phenomenon have also arisen, which is why several organizations and States have presented regulatory instruments establishing ethical and legal contours for the development and application of this technology in society. In Brazil, until the closing date defined for this work, the Bills No. 5051/2019, 5691/2019, 21/2020, 240/2020, 4120/2020, 872/2021, 1969/2021 e 705/2022 were presented in the Chamber of Deputies and in the Federal Senate, with the aim of regulating artificial intelligence in the country. The purpose of the present work is, through a qualitative approach, inductive reasoning and content analysis procedure of a legal-social aspect and of a legal-interpretative and legal-propositional type, to analyze the provisions of these bills in the face of the specialized literature and the results achieved in the application of a multilayer neural network, in order to decompose and examine the main aspects concerning the regulation of artificial intelligence, with a focus on the national context, which, due to the recency of the legislative proposals, still does not find a variety of specific analyses. In chapter I, legal norms in the scope of innovation are contextualized, with the mapping of existing norms in Brazil, at the federal level, regarding artificial intelligence. Chapter II, in turn, explains, initially, the Brazilian legislative process and then the main provisions of the mentioned bills, as well as the stage they are in the legislative process. In chapter III, the application of a multilayer neural network to the labor judicial area using the Julia programming language is presented, while in chapter IV the issues of the bills referring to the concept of artificial intelligence, explainability and transparency, reliability and security, personal data protection, responsibility and compliance with fundamental rights and guarantees are analyzed in the face of the specialized literature and the results identified in the application of the presented neural network. The final considerations, containing the results obtained through the research carried out, conclude this work, which is expected to be able to subsidize and foster the regulatory discussions underway in Brazil regarding artificial intelligence.