The Use of Artificial Intelligence Systems in Law Enforcement

Abstract

The paper examines the technological basis and opportunities for the use of artificial intelligence systems in law enforcement. The authors describe the investigation methods and the essence of artificial intelligence, and conduct a detailed study of approaches to the taxonomy of its systems. Artificial intelligence today does not only make it possible to solve specific tasks, but also approaches human cognition in its functions. In the present legal environment, programming and automation of crime investigation and solving are used to create information and reference systems, as well as databases and criminalistic algorithms that optimize, for example, the process of developing and verifying criminalistic leads, planning an investigation, supporting the maintenance of order, searching for the culprit, etc. The authors define key features of artificial neural networks viewed as one of the main methods of using artificial intelligence systems in law enforcement, specifically, situational adaptive learning, ability to identify non-obvious links and regularities. The designing of an applied artificial neural network is examined stage-by-stage. At the first stage, a dataset is collected - it is a volume of data for training the network. At the second stage, an algorithm (a set of rules) for learning is selected or designed. After that comes the process of learning and validating its results. The authors analyze the criteria for evaluating the effectiveness of training an artificial intelligence system, including the criteria of precision and accuracy. They single out three key types of operations in the sphere of law enforcement that can be performed by artificial intelligence systems: identification (of visual images and links between the objects of criminalistic study), prediction and classification.