Application of Artificial Intelligence Technology to Assess the Correction of Convicts on Early Release from Punishment

Abstract

Artificial intelligence technology is increasingly penetrating into various spheres of modern life and causes heated discussions about the possibility of its use in the administration of justice. The paper analyzes the positions of supporters and opponents of the use of artificial intelligence directly in the role of a judge and especially when making evaluative decisions when resolving issues of conditional early release or replacing the unserved part of the sentence with a milder type of punishment. Analysis of the results testing the GigaChat neural network showed that artificial intelligence is not able to take responsibility when deciding whether to grant or refuse to grant parole requests, but it is capable of structuring positive and negative criteria that characterize the behavior of the convict, and which can influence the court’s decision. Artificial intelligence, comparing the professionally significant qualities of a human judge and an AI judge, gives preference to a human, since he has unique characteristics that are not typical for technology.