On approaches to using neural networks as an object and a means of learning in primary and secondary school

Problem statement. The article analyzes scientific papers on the application of artificial intelligence in various fields of human activity and the use of neural networks in education. The problem that requires research is that such networks should be considered not only as an object for study, but also as a means of teaching schoolchildren. Methodology. In the course of the research, the author’s model of integrated use of neural networks as an object and a means of learning has been developed and tested. To determine the truth of the statements put forward, a pedagogical experiment based on the use of Pearson’s criterion χ2 was conducted in computer science classes of the school No. 293 in Moscow. Results. The paper shows the reasons for rapid development of neural networks. The methods and examples of their application in computer science lessons in primary and secondary school as well as examples of generating unreliable material are proposed. The examples of programs and neural networks for schoolchildren to work with multimedia materials are given. The stages of schoolchildren’s work are proposed using the example of the Kandinsky 3.1 neural network. The results of an experimental test of the effectiveness of using neural networks in preparing schoolchildren to work with the content of electronic publications are described. Conclusion. For the effective use of neural networks in teaching, it is recommended to consider them as an object of study, paying attention to the problem of reliability of generated material, threats and risks of constant use of neural networks by schoolchildren. When using such networks as a learning tool, it is necessary to explain the criteria on the basis of which a student will choose a neural network depending on existing tasks and interests.