Russian students on the potential and limitations of artificial intelligence in education

The rapid entry of artificial intelligence (hereinafter AI) into all spheres of social life determines the need to consider ongoing changes and to conduct systematic sociological research. Education and science are key resources that, on the one hand, develop and improve AI technologies, but, on the other hand, fully experience the pressure of contradictions caused by new technologies. It is important for both the higher education system and society as a whole to understand how students react to new opportunities and technologies, how involved they are in the use of AI in their educational activities, and how they evaluate their experience of applying new technologies in learning. The article presents data on the Russian university students’ assessment of their personal experience of using generative AI models (neural networks) in educational activities, highlights the most popular AI functions, and evaluates students’ satisfaction with their use. The article is based on the survey of Russian university students conducted in 2023-2024 (N = 52919), which showed that, despite the current massive fascination with digital technologies and the use of neural networks, Russian students assess quite ambiguously their use in studies, and in senior years, this assessment becomes more critical and balanced (concerning the opportunities provided by AI). The survey data allows to conclude that the use of generative AI models in education requires a set of decisions on the direct regulation of its application and ethical issues, the thorough revision of the students’ forms of independent work, including final certifications and test tasks, and a search for constructive approaches to the use of AI to improve the quality of education and the work of the higher education system. Moreover, AI assigns the higher education system a task of developing students’ critical assessment of the results of interaction between a human being and a neural network, focusing on the limitations and capabilities of the generated information, and acceptable formats for its use in research and learning.