The Use of Neural Network Language Models in Judicial Investigation in Criminal Proceedings

The article explores the challenges and prospects of using generative artificial intelligence within the framework of forensic evidence. The conclusion is justified that the neural network is not capable of acting as a subject for evaluating evidence, but it should to create a text model of the judicial investigation and its subsequent analysis for the presence of logical contradictions, inconsistencies and other “cognitive” shortcomings. The results obtained should used by the court in evidence, which will serve to strengthen its activity when considering a criminal case on the merits, to improve the quality of the administration of justice in general.