Poder artificial de tributar? Limites e requisitos à utilização (adequada) da inteligência artificial pela administração tributária

Abstract

This thesis aims to analyze the limits regarding the use of Artificial Intelligence by Tax Administration, as well as the requirements that must be respected by it, as a condition for its use to be considered adequate in the light of the constitutional text and the concept of a Democratic State. It argues that the minimum framework of these limits and requirements can be formulated from a systemic and teleological interpretation of the Federal Constitution, the National Tax Code and of the provisions of the ordinary legislation; especially when this interpretation is associated with an epistemological appreciation of the challenges arising from the management of Artificial Intelligence. By addressing issues that become increasingly delicate as the use of Artificial Intelligence is intensified by the State – such as the need to guarantee transparency and the right of defense and privacy – this research argues that the use of computer systems reshapes the dynamics of the relationship between tax and law, whether it is to demand that new duties are imposed on the Tax Administration or to demand that traditionally established duties are reformulated in light of this new scenario. On top of that, this thesis also proposes, based on the principles developed in this study, that a list of requirements/duties should be fulfilled by the State whenever this type of technology is used. Although it is not an exhaustive and static list, which would even be incompatible with the disruptive nature of this technology, this approach allows the construction of parameters, even if they are initial ones, which, if fulfilled, would minimize the risk of violation of fundamental rights of the citizen/taxpayer. This systematization also makes it possible to identify what the State should be concerned with as it accelerates, sometimes even carelessly, the use of intelligent computational models, helping the development of new research on the subject. In its conclusion, the thesis presents a critical analysis of the legislation in progress in the National Congress that aims to promote the regulation of the life cycle of Artificial Intelligence, seeking not only to point out the advances obtained so far, but also to highlight the bottlenecks which need to be managed. It also alerts to the absence of specific provisions aimed at regulating the use of Artificial Intelligence by the Public Administration, justifying the need for its insertion. In the end, a proposal for a legislative writing is formulated and that implements deeper, clear and concrete state duties, especially with regard to the principle of transparency; seeking to enable a more balanced scenario, in which the purposes of combating illegality and increasing tax collection are achieved without promoting the unnecessary weakening of fundamental rights which are essential to the maintenance of the notion of the Rule of Law. This is an exploratory thesis with a prospective nature. In termos of procedure, the thesis has adopted bibliographical and documental research in addition to case analysis, using the technique of indirect documental data collection. The selected data are analyzed in a quali-quantitative perspective, from which possible solutions to the established research problema are formulated.