Abstract:

This contribution offers a critical synthesis that examines how generative AI (genAI) has been framed since ChatGPT's public release in late 2022 and contrasts this framing with established findings from educational technology research. Drawing on Clark's (1994) media debate, cognitive load theory (Sweller et al., 2011), generative learning research (Fiorella & Mayer, 2016), and recent methodological critiques of ChatGPT studies (Weidlich et al., 2025), six theses are derived:
(1) GenAI has not overturned the fundamental mechanisms of human learning; perception–action cycles, multiple memory systems, prior knowledge, and learner activity remain the primary drivers of durable learning, while genAI mainly alters the informational environment in which these mechanisms operate.
(2) GenAI does not have an intrinsic learning effect; where methodologically credible, reported gains are better explained by changes in task design, practice opportunities, and feedback structures than by the medium itself.
(3) Media-comparison studies that contrast "with AI" versus "without AI" conditions have limited explanatory value because they typically conflate technology, pedagogy, and assessment; interpretable evidence requires designs that specify the instructional function of genAI rather than treating it as a black-box treatment.
(4) When genAI is deliberately aligned with evidence-informed principles of generative learning, it can strengthen practices already known to be effective; by contrast, outsourcing cognitively demanding tasks to genAI risks "deepfake learning" (Hodges, 2025), in which ostensibly generative tasks become cognitively passive.
(5) Learners often behave as discipuli economici (Hodges & Kirschner, 2024): as long as assessment regimes reward polished products over learning processes, students will rationally minimise effort, shifting the design problem from policing tools to redesigning tasks and criteria.
(6) GenAI does not determine educational practice; institutions retain responsibility for defining purposes, boundaries, and conditions of use. Taken together, these theses suggest that the central lesson from genAI is not a technological revolution in learning, but the need to recentre learning theory, task design, and assessment in how we research and debate its role in higher education.

References:
[1] Clark, R. E. (1994). Media will never influence learning. Educational Technology Research and Development, 42(2), 21–29. https://doi.org/10.1007/BF02299088
[2] Fiorella, L., & Mayer, R. E. (2016). Eight ways to promote generative learning. Educational Psychology Review, 28(4), 717–741. https://doi.org/10.1007/s10648-015-9348-9
[3] Hodges, C. B. (2025). Deepfake learning: Assessing learning in the generative AI era. In S. Papadakis (Ed.), AI roles and responsibilities in education (pp. 185–206). Springer. https://doi.org/10.1007/978-3-031-96855-6_13
[4] Hodges, C. B., & Kirschner, P. A. (2024). Innovation of instructional design and assessment in the age of generative artificial intelligence. TechTrends, 68(1), 195–199. https://doi.org/10.1007/s11528-023-00926-x
[5] Sweller, J., Ayres, P., & Kalyuga, S. (2011). Cognitive load theory. Springer. https://doi.org/10.1007/978-1-4419-8126-4
[6] Weidlich, J., Gašević, D., Drachsler, H., & Kirschner, P. A. (2025). ChatGPT in education: An effect in search of a cause. Journal of Computer Assisted Learning, 41, e70105. https://doi.org/10.1111/jcal.70105

Keywords:

Educational technology, genAI, artificial intelligence, learning, generative learning, ChatGPT, higher education.