How Can Behavioural and Psychological Theories Explain Patterns of Artificial Intelligence (AI) Use Among 15-Year-Old Students, and How Are These Patterns Associated With Learning Outcomes and Academic Performance?

Generative artificial intelligence (AI) provides students with rapid access to explanations, summaries, and suggested answers. For 15-year-old learners, this accessibility raises an important educational and behavioural question: under what circumstances does AI support meaningful engagement with challenging learning materials, and when might it reduce opportunities for independent cognitive effort? This paper reviews published research in behavioural science and selected findings from the OECD's PISA 2025 Results (Volume I), with particular emphasis on Uzbekistan. Cognitive-demand avoidance and cognitive offloading are employed as theoretical frameworks for interpreting patterns of AI use, while a clear distinction is maintained between these concepts and the variables directly measured by PISA. According to the OECD, 48.6% of students in Uzbekistan use AI chatbots at least once a week to support their learning, compared with the OECD average of 45.5%. Furthermore, 80.6% of students in Uzbekistan report learning at school how to evaluate AI-generated information, compared with an OECD average of 62.6% (OECD, 2026). These findings indicate that AI-assisted learning is already widespread among students. However, the reported figures do not establish how frequently students substitute AI-generated summaries for assigned reading materials or identify the motivations underlying individual decisions to use AI in this way. The paper argues that the educational implications of AI use depend on the nature of the cognitive tasks delegated to AI, the learner's objectives, and the extent to which AI-generated information is critically evaluated and integrated into independent learning. As the study relies on secondary sources and does not involve the analysis of individual-level student records or the collection of primary empirical data, the proposed explanations should be understood as theoretically informed interpretations rather than established causal relationships. The paper concludes that educational institutions should develop students' ability not only to use AI tools effectively but also to recognise when independent reading, source verification, and critical reasoning remain essential for meaningful learning.