Regulação da inteligência artificial à luz dos desafios impostos pela tecnologia à responsabilidade civil

Abstract

This study proposes to investigate the challenges imposed by machine learning systems on the foundations of tort law in Brazil, from the perspective of the technology’s different applications and degrees of complexity, and, therefore, the impossibility of a generic definition of a single liability regime (strict or negligence-based) for damages caused by the emergent behavior of artificial intelligence systems. Faced with this impossibility, it is found that the establishment of a principle-based regulatory regime, surrounded by normative microsystems of each area of application of technology, when necessary, would be welcome in the current stage of development of AI. In order to identify the main principles related to liability and AI on the global stage, the main regulations on artificial intelligence in the United States, China, United Kingdom and Brazil are studied. From this analysis, it is identified that accountability, explainability, and transparency are essential principles to regulate the AI liability issue. Given this information, a parallel is drawn between these principles, tort law in Brazil, and its application to cases of damage caused by artificial intelligence.