Integrating artificial intelligence into higher education balances student learning benefits and cognitive overload

Artificial intelligence (AI) is increasingly integrated into higher education, reshaping how students access information, complete academic tasks, and engage with learning. Despite its growing educational role, the implications of AI-supported learning for students’ psychological well-being remain insufficiently understood, particularly in developing-country contexts. This study examines how Perceived Usefulness of AI Tools in Learning (PUAITL) and Frequency of AI Tool Usage in Academic Activities (FAITUAA) are associated with Student Mental Health (SMH), while investigating Cognitive Overload (CO) as a mediating mechanism and Digital Self-Efficacy (DSE) as a moderating factor.

Data were collected through a cross-sectional survey of 390 undergraduate and postgraduate students from public and private universities in Bangladesh and analyzed using PLS-SEM with Smart-PLS 4.0 and bootstrapping procedures. The findings indicate that both perceived usefulness and frequency of academic AI use are positively associated with SMH. Importantly, both PUAITL and FAITUAA are associated with lower cognitive overload, whereas greater cognitive overload is associated with poorer student mental health.

CO partially mediates both relationships, indicating that AI-supported learning is associated with psychological well-being both directly and indirectly through reduced cognitive burden. DSE significantly strengthens the positive relationship between PUAITL and SMH, whereas its moderating effect on the relationship between FAITUAA and SMH is not statistically significant. Drawing on Cognitive Load Theory, Conservation of Resources Theory, and Technostress Theory, the study provides an integrated explanation of the cognitive, resource-related, and psychological mechanisms underlying AI-supported learning.

The findings emphasize the importance of responsible AI integration, cognitive-load management, digital capability development, and context-sensitive institutional support for promoting psychologically sustainable AI-enabled learning environments in higher education. This research was supported by the University of Debrecen Program for Scientific Publication. This research was funded by Princess Nourah bint Abdulrahman University Researchers Supporting Project number (PNURSP2026R797), Princess Nourah bint Abdulrahman University, Riyadh, Saudi Arabia.

Department of Management, Faculty of Business Administration, University of Tabuk, Tabuk, Saudi Arabia Vice Dean of Postgraduate and Development, Department of Management, Faculty of Business Administration, University of Tabuk, Tabuk, Saudi Arabia Department of Finance, College of Business Administration, Prince Mohammad Bin Fahd University, Al Khobar, Eastern Province, Saudi Arabia Center for Sustainable Business and Innovation, Prince Mohammad Bin Fahd University, Al Khobar, Eastern Province, Saudi Arabia Coordination and Research Centre for Social Sciences, Faculty of Economics and Business, University of Debrecen, Böszörményi Street 138, Debrecen, 4032, Hungary Department of Business Administration, Faculty of Business Studies, Bangladesh Army University of Science and Technology, Saidpur, Nilphamari, 5310, Bangladesh Fakulti Sains dan Matematik, Universiti Pendidikan Sultan Idris, Tanjung Malim, Malaysia Department of Management, College of Business Administration, Princess Nourah bint Abdulrahman University, P.O. Box 84428, 11671, Riyadh, Saudi Arabia College of Business Administration, Prince Mohammad Bin Fahd University, Al Khobar, Eastern Province, Saudi Arabia Department of Business Administration, Northern University Bangladesh, 111/2 Kawlar Jame Mosjid Road, Dakhsinkhan, Ashkona-1230, Dhaka, Bangladesh Correspondence to Md Aminul Islam, Mohammad Bin Amin or Mohammad Abul Kalam Azad. The authors declare no competing interests.

During the preparation of this manuscript, the authors utilized Grammarly solely to improve the clarity, grammar, and overall quality of the English language because English is not their native language. All content generated for the manuscript was critically reviewed, revised, and validated by the authors after using the software. The authors assume full responsibility for the accuracy, integrity, originality, and final presentation of the manuscript, as well as all interpretations and conclusions reported in this study.