Inteligência artificial, discriminação algorítmica e o dever de diligência dos administradores das sociedades anônimas

Abstract

The research analyzes the state of the art of algorithmic discrimination reproduced by some tools of artificial intelligence in order to, from the current factual reality, interpret what should be the content of the duty of care of company directors, provided for in Article 153 of Law 6.404/76, identifying whether there is the emergence of duties arising from these new potential damages. The master's dissertation follows the deductive method, reaching the conclusions pursued from two general premises (legal impacts of algorithmic discrimination and duty of care), to indicate the existence of a more specific interpretation of the duty of care applied to artificial intelligence. The verified hypothesis assumes that the administrator's duty of care can be interpreted in a way that requires caution in decision-making involving the use of artificial intelligence, especially those used in decision-making processes that offer (or deny) services, goods and rights to individuals and may be discriminatory. The text begins with a brief contextualization of the expansion of artificial intelligence, seeking to deepen the understanding of the current state of the art and, more specifically, what algorithmic discrimination means. Next, the research is dedicated to clarifying the concepts of artificial intelligence, algorithms, and machine learning, which are relevant to the understanding of algorithmic discrimination and its causes. The second chapter examines what diligence is expected of a 21st century administrator, especially one who operates in the year 2023, considering the fact that, since the enactment of Law 6.404/76, with article 153, almost five decades have passed and access to information and data, which can subsidize business decisions, has changed a lot. The research analyzes the evolution of the interpretation that the doctrine of corporate law gives to article 153 of Law 6.404/76, as well as the interpretation given by the Securities and Exchange Commission of Brazil (Comissão de Valores Mobiliários) - which will be done based on emblematic examples, given the large volume of precedents that address the duty of care - in order to, based on this study, list the duties that the administrator of a company should adopt in any business decision. The third chapter seeks to unite the two investigations in order to interpret what actions the managers of companies - developers or users of artificial intelligence in their decisions and business processes - should adopt in order to avoid and/or mitigate illegalities or unconstitutionalities that arise from their applications of artificial intelligence. Or even if they do occur, this chapter seeks to understand how to prove that such managers have been zealous in adopting all possible measures to avoid these problems - all this in light of the most recent interpretation of the practical command contained in article 153 of Law 6.404/76. The conclusion of the research is in the sense of identifying that artificial intelligence arises together with the fulfillment of a new duty of care: the duty to implement compliance programs and governance structures apt to mitigate or correct algorithmic discrimination.