Racial segregation and algorithmic biases: racism machines in the context of criminal control

Abstract

This paper aims to verify if the use of Artificial Intelligence Systems could reduce the very high load of Criminal Control’s selectivity, especially those based on racism and discrimination. Initially, at first, for this purpose, the work deals with the selectivity of punitive power as a structural characteristic. Subsequently, we carefully observe the data that are used in the construction of AI models. After that, it lists possible solutions to reduce the incidence of biased algorithms and mitigate the reproduction of racism through the use of the machine. Finally, we propose to identify some negative externalities resulting from non-observance of certain ethical standards in AI models. The research is justified in view of the need to know the perverse reflexes that biased algorithms can produce in the Criminal Control segments, catalyzing prejudices and segregating social sectors chosen by skin color. The research method is the inductive one, supported by bibliographic research and official numbers extracted from official databases.