Abstract
<jats:p>With the constant development of computer technology and the increasing complexity of software, the issues of efficient generation of program code are becoming more and more important. The use of modern models and methods of artificial intelligence (AI) to automate the process of creating program code is one of the promising areas of research in the field of software engineering. With the advent and development of neural networks and deep learning, the possibilities for automating the process of creating program code have expanded significantly. In addition, the growing complexity of software and the increased requirements for its quality and performance emphasize the relevance of researching the possibilities of using modern AI methods to generate program code. The purpose of the work is to analyze the quality of application of AI models in the generation of program code intended for the development of complex software systems by students of the “Software Engineering” training direction. In the course of the study, the subject area related to the application of the capabilities of modern AI models to generate program code was studied. Various approaches and technologies of AI were considered, as well as their use in the training of future software engineers. The study included the organization of queries to the ChatGPT-4o, Copilot, Google-gemini-1.5-pro, and YandexGPT models and subsequent comparative analysis of the software codes generated by these models. For sending and subsequent processing of requests, an automated client-server application was developed, which was used in the study of the potential of AI models for generating program code. The results obtained can be useful for teachers and students of the direction of training “Software Engineering”, as well as for software developers using AI models to automate programming processes and improve the quality of professional training.</jats:p>