The latest AI technologies in the educational environment: systematization of research and applied aspects

Abstract

The article systematizes modern approaches to implementing AI technologies in higher education, outlining their didactic functions, institutional prerequisites, and economic feasibility in the training of students in physics, mathematics, and technical specialties. An integrated approach is presented, which involves a combination of intelligent tutoring systems, generative language models and educational analytics tools in conjunction with formative assessment and the «person-in-the-cycle» principle. The purpose of this study is to systematize scientific approaches and applied practices of utilizing AI technologies in higher education and to determine their potential for enhancing learning outcomes, provided that ethical and legal standards are observed. Methods. A review of the current literature, a comparative analysis of international and national cases, a typology of tools based on learning and assessment functions, and an economic justification for a basic pilot using available licenses and standardized performance metrics are carried out. Results. The main classes of educational AI technologies (Intelligent Tutoring Systems, Large Language Models, Educational Analytic Systems, Adaptive Assessment Systems, AI-driven Content Generation, AI assistants for teachers) and their pedagogical effects are identified; key risks are identified (psycho-emotional disorders, model bias, violation of the principles of academic integrity, threats to data privacy) and the conditions for their minimization are specified (application of the «person-in-the-loop» approach, development of transparent rules of use, content localization, implementation of data management policies). The generalized empirical data indicate a decrease in the frequency of typical errors, an increase in the level of assimilation of theoretical material, a reduction in the time required to complete practical tasks, and an improvement in digital competencies. The financial feasibility of the initial implementation for institutions with limited resources has been proven. Conclusions. The systematic application of AI technologies in the educational process, in combination with formative assessment, the use of localized content, and the training of teaching staff, is pedagogically feasible. At the same time, the effectiveness of the implementation requires compliance with ethical policies, transparent rules for ensuring academic integrity, and long-term monitoring of results.