Features of Plagiarism Checking in a Technical University Considering the Possibilities of Application of Generative Artificial Intelligence by Learners

The purpose of the study . The paper discusses the current issues of checking the volume of borrowed text in the graduation qualification paper in a technical university taking into account the probability of using the capabilities of artificial intelligence by students. The problem of plagiarism, in particular plagiarism of graduation qualification papers (diploma theses), has always been actual. Some students, when writing their graduation qualification papers, tend to copy texts of papers defended in previous years, which led to the need to organize plagiarism check of texts of all papers of the current year of graduation. There are various methods that make it possible to easily bypass such a check. This problem has become especially actual in recent years due to the development of information technology. The widespread introduction of generative artificial intelligence has led to the emergence of a new problem the need for the supervisor and/or designated responsible person to check the graduation qualification paper for the presence of text generated by artificial intelligence. This paper discusses the features of plagiarism check of texts of graduation qualification papers of students studying in the areas related to information technologies, considering the potential possibility of using generative artificial intelligence by students, in particular ChatGPT and GitHub Copilot software. Materials and methods . The method of comparative analysis of scientific publications for plagiarism checking and the use of artificial intelligence in the educational process was used. Existing plagiarism checking methods are irrelevant when checking texts generated by artificial intelligence. The attributes and examples of such texts are considered. Trends in the environment of students at a technical university in relation to the use of generative artificial intelligence, in particular ChatGPT and GitHub Copilot software when writing graduation qualification papers were experimentally identified. The possibilities of applying a number of programs for detecting texts generated by artificial intelligence have been verified. Research results . The analysis of plagiarism check results for texts generated by an artificial intelligence system and prepared by a methodologist was carried out. The problem of unambiguous automatic detection of the use of generative neural networks by students in the process of preparing a graduation qualification paper due to the presence of false positives was discussed. It seems advisable to widely implement systems for checking the text of graduation qualification paper for the presence of text generated by artificial intelligence systems. However, the test use of existing verification systems showed that the reliability of checking for the presence of text generated by artificial intelligence systems is highly debatable. The percentage of identified borrowings can vary both downwards and upwards with incorrect conclusions. The problems caused by the peculiarities of teaching students at a technical university are discussed. A path for checking the materials of graduation thesis for AI plagiarism is proposed. Conclusion . The importance and necessity of checking the originality of the graduation qualification papers for borrowings both the texts of the graduation qualification paper of previous years and the use of texts and programs generated by artificial intelligence systems are outlined. The authors propose possible approaches to organizing the educational process at a technical university taking into account the accumulated experience, as well as ways to solve the problems discussed in the paper, in particular, the introduction of mandatory marking of both the text and the program code created by the artificial intelligence system is proposed. In addition, the need to develop relevant teaching methods, including the formation of reflexivity, is emphasized.