Implementation of artificial intelligence in the system for detecting academic dishonesty in Ukrainian secondary and higher education institutions

Abstract

Artificial intelligence (AI) systems in education and science require special attention due to the rapid development of digitalization. The article aims to determine the effectiveness of modern AI systems based on a comparative analysis of AI methods and to develop scientifically sound prospects for their widespread implementation in Ukrainian education. The author applied the PRISMA scientific approach to achieve the proposed goal, which allowed the selection of the necessary scientific sources (50) and their systematization for further analysis. The results show that natural language processing, latent semantic analysis, word embeddings, stylometry analysis, text analysis and separation, graph methods, and data integrity checks are used to check for dishonesty. These methods are also used to determine the authorship of a text, identify suspicious moments in texts, and detect plagiarism and borrowings. Other modern programs allow you to identify the facts of academic dishonesty and plagiarism, even when the text is paraphrased. The main problem is the possibility of circumventing the main algorithms by changing the structure of the text or generating false positive results. For the modern educational system of Ukraine, it is proposed that a high-quality, clear state strategy be formed, teachers' digital literacy be developed, AI tools be introduced into teaching and teacher training methods, and teachers be trained to use platforms for automated assessment and personalization of learning. The conclusions indicate that this issue will require further updating due to the development of technology.