This article is devoted to the study of the possibility of developing effective exercises in object-oriented programming using ChatGPT, which would provide a deep understanding and application of OOP concepts, as well as determining the technological stages of the process of their generation. Using OpenAI Copilot as a large language model, we create programming exercises (including sample solutions and test cases), evaluating them qualitatively and quantitatively. Our results show that most of the automatically generated content is both new and correctly formulated, and in some cases completely ready for use. To generate exercises, primings were developed as input data for Copilot. They are templates describing the concepts of object-oriented programming, data structures and keywords for the formulation of task conditions. Our analysis shows that mass models of generative machine learning are of considerable value as a tool for teachers, although there is still a need for some control to ensure the quality of generated content before it is provided to students.