The ethical dilemma of algorithmic decision in Public Administration

Abstract

The use of artificial intelligence to support decision-making is a reality in the current scenario. Discussions on data processing – whether through supervised learning or through unsupervised learning (deep learning) – gain relevance in the field of law. At the same time that they gain relevance, however, challenges arise regarding the results that emerge due to data processing (output), insofar as these may be biased due to the data set coming from the physical world, of conceptions and human relations. Due to these challenges, this article aims to point out such aspects in the contemporary context and propose measures that can be adopted as possible solutions. The analysis about the existence of an antidote as a counterpoint to the bias is necessary, in order to qualify the continuous and unstoppable use of these technologies. Dealing with the difficulties of the subject in the constitutional scope and in the light of Brazilian legislation, in turn, justifies the importance of the subject and allowed a proposal for such challenges that arise due to the use of artificial intelligence. Therefore, this article makes use of bibliographical research on the subject and of systemic interpretation between the Federal Constitution and legislation that address legal aspects of Public Administration.