What impact will large language models like ChatGPT have on higher education? The optimists or “AI bloomers” claim that AI tools will usher in a new era of personalized learning and allow instructors to offload grading and other tedious and time-consuming tasks to ChatGPT, thus freeing up more time for them to devote to their students. Representative of this camp is this paper by Miguel Ángel Escotet, the rector of the Intercontinental University of Business, who predicts that AI-powered “intelligent tutoring systems” (ITSs) will “provide “personalized, adaptive instruction and student feedback based on their assessments and progress.” In other words, ChatGPT will not only meet students where they are; it will make them smarter. (See also Suriano et al. 2025, available here.)

Critics or “AI doomers”, by contrast, contend that ChatGPT poses a dire threat to the central mission of higher ed: the promotion of critical thinking. Representative of this opposing camp is this paper by Michael Gerlich, the director of Center for Strategic Corporate Foresight and Sustainability at the SBS Swiss Business School, who surveyed and interviewed over 600 participants and found “a significant negative correlation between frequent AI tool usage and critical thinking abilities, mediated by increased cognitive offloading.” In plain English, when students turn to ChatGPT for ready-made answers, they are bypassing the mental effort to engage in critical thinking, i.e. the ability to analyze, evaluate, and synthesize information. (See also Lee et al. 2025, available here and here.)

Alas, all of these papers — like most, if not all, academic research on education — are woefully inadequate. Escotet’s paper is just a series of personal reflections, while Gerlich’s findings, as well as those of Lee et al. 2025 and Suriano et al. 2025, are based on surveys and self-reports. But personal reflections (no matter how respected the author) or a mere survey (no matter how many people are interviewed) will not do. In order to draw any reliable or rigorous conclusions about ChatGPT’s impact on higher ed, we would need to conduct a properly-designed experiment, and we would also need to assign students to two different groups: a control group and a treatment group.

The problem, however, is that that type of experimental research usually requires pre-approval by an institutional review board (IRB), and it is unlikely that an IRB would approve experimental research on students directly, especially when a treatment is expected to benefit one group of students over another group of students. (Cf. Rebele & St. Pierre 2015, p. 133 n.3.) So, what is to be done? How can we really measure or “test” the impact of AI tools on higher ed without establishing a control group (students who are not allowed to use AI) separate from the treatment group (students who are allowed to use AI)?

For my part, I will remain agnostic (for now) about the relative merits of these opposing camps — the AI doomers and bloomers — until such an experimental study (or better yet, a series of such studies) is completed. In the meantime, I will take a step back from this great debate in order to address an important preliminary question: what do we mean by “critical thinking” in the first place? After all, before we can even begin to measure something as vague and ambiguous as “critical thinking”, we need to define what we mean by this term. I will take a first stab at a definition in my next post.

Works cited

Escotet, M. A. (2023) The optimistic future of Artificial Intelligence in higher education. Prospects, 54(3), 531-540. https://doi.org/10.1007/s11125-023-09642-z

Gerlich, M. (2025) AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking. Societies, 15(1), 6. https://doi.org/10.3390/soc15010006

Lee, H. P., Sarkar, A., Tankelevitch, L., Drosos, I., Rintel, S., Banks, R., & Wilson, N. (2025) The impact of generative AI on critical thinking: Self-reported reductions in cognitive effort and confidence effects from a survey of knowledge workers. Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems (pp. 1-22). https://dl.acm.org/doi/10.1145/3706598.3713778

Rebele, J. E., & Pierre, E. K. S. (2015) Stagnation in accounting education research. Journal of Accounting Education, 33(2), 128-137. https://doi.org/10.1016/j.jaccedu.2015.04.003

Suriano, R., Plebe, A., Acciai, A., & Fabio, R. A. (2025) Student interaction with ChatGPT can promote complex critical thinking skills. Learning and Instruction, 95, 102011. https://doi.org/10.1016/j.learninstruc.2024.102011