Keywords
metacognition
self-regulated learning
calibration
cognitive offloading
higher education
experimental study 
Abstract
The rapid integration of generative artificial intelligence (AI) tools in higher education has prompted significant scholarly debate regarding their pedagogical implications. While these tools offer unprecedented opportunities for personalized learning support, concerns have been raised about their potential to undermine students' metacognitive development by enabling cognitive offloading and fostering intellectual dependency.
This experimental study investigated the causal impact of generative AI tutors on undergraduate students' metacognitive monitoring accuracy (calibration) and self-regulated learning processes.
One hundred and thirty-eight (138) undergraduate students were randomly assigned to either an experimental condition, in which they completed complex problem-solving tasks in introductory programming with access to a generative AI tutor, or a control condition, in which they used traditional instructional resources. Metacognitive monitoring accuracy was measured through judgments of learning and compared with actual performance. Interaction logs and think-aloud protocols were analyzed to examine self-regulatory processes. A transfer task completed without AI support assessed the durability of observed effects.
Students who used the AI tutor demonstrated significantly poorer calibration accuracy compared to the control group (p < .001, d = 0.75). Interaction log analysis revealed that AI users predominantly engaged in direct answer requests (44%) and verification requests (32%), indicating substantial cognitive offloading. Prior knowledge moderated the effect, with lower-knowledge students being most adversely affected. The metacognitive deficit persisted on a transfer task completed without AI support (p < .001, d = 0.83).
This study provides the first rigorous experimental evidence that generative AI tutors can impair metacognitive monitoring, particularly among novice learners. The findings extend theories of self-regulated learning to account for distributed cognition with AI tools and have significant implications for instructional design, AI literacy curricula, and institutional policies governing AI use in higher education.
As generative AI becomes increasingly prevalent in educational settings, these findings underscore the urgent need for pedagogical frameworks that harness the benefits of AI while safeguarding the development of students' metacognitive autonomy and capacity for lifelong learning.
References
1. Aleven, V., & Koedinger, K. R. (2002). An effective metacognitive strategy: Learning by doing and explaining with a computer-based Cognitive Tutor. Cognitive Science, *26*(2), 147-179.
2. Aleven, V., Stahl, E., Schworm, S., Fischer, F., & Wallace, R. (2003). Help seeking and help design in interactive learning environments. Review of Educational Research, *73*(3), 277-320.
3. Azevedo, R., & Cromley, J. G. (2004). Does training on self-regulated learning facilitate students' learning with hypermedia? Journal of Educational Psychology, *96*(3), 523-535.
4. Azevedo, R., & Gašević, D. (2019). Analyzing multimodal multichannel data about self-regulated learning with advanced learning technologies: A call for action. Journal of Learning Analytics, *6*(3), 1-10.
5. Bond, M., Khosravi, H., De Laat, M., Bergdahl, N., Negrea, V., Oxley, E., ... & Siemens, G. (2024). A meta-systematic review of artificial intelligence in higher education: A call for increased ethics, collaboration and rigour. Studies in Higher Education, 1-23. Advance online publication.
6. Clark, R. E. (2020). Learning from media: Arguments, analysis, and evidence (2nd ed.). Information Age Publishing.
7. Creswell, J. W., & Creswell, J. D. (2018). Research design: Qualitative, quantitative, and mixed methods approaches (5th ed.). SAGE Publications.
8. Dunlosky, J., & Thiede, K. W. (2013). The role of metacognition in learning. In D. Reisberg (Ed.), The Oxford Handbook of Cognitive Psychology (pp. 567-581). Oxford University Press.
9. Ericsson, K. A., & Simon, H. A. (1993). Protocol analysis: Verbal reports as data (Rev. ed.). MIT Press.
10. Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, *39*(2), 175-191.
11. Flavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of
cognitive–developmental inquiry. American Psychologist, *34*(10), 906-911.
12. Jager, J., Putnick, D. L., & Bornstein, M. H. (2017). II. More than just convenient: The scientific merits of homogeneous convenience samples. Monographs of the Society for Research in Child Development, *82*(2), 13-30.
13. Johnson, M., & Smith, R. (2023). Student and faculty perceptions of generative AI in higher education: A mixed-methods study. The Internet and Higher Education, *59*, Article 100921.
14. Kalyuga, S. (2007). Expertise reversal effect and its implications for learner-tailored instruction. Educational Psychology Review, *19*(4), 509-539.
15. Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic inquiry. SAGE Publications.
16. Luckin, R., Cukurova, M., Kent, C., & du Boulay, B. (2022). Empowering educators to be AI-ready. Computers and Education: Artificial Intelligence, *3*, Article 100076.
17. Molenaar, I. (2022). The concept of hybrid human-AI regulation: An agenda for learning analytics research. Journal of Learning Analytics, *9*(2), 1-14.
18. Nelson, T. O., & Narens, L. (1990). Metamemory: A theoretical framework and new findings. In G. H. Bower (Ed.), The psychology of learning and motivation (Vol. 26, pp. 125-141). Academic Press.
19. Risko, E. F., & Gilbert, S. J. (2016). Cognitive offloading. Trends in Cognitive Sciences, *20*(9), 676-688.
20. Schraw, G. (2009). A conceptual analysis of five measures of metacognitive monitoring. Metacognition and Learning, *4*(1), 33-45.
21. Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-experimental designs for generalized causal inference. Houghton Mifflin.
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
Copyright (c) 2026 Authors