Rebecca A. Huggins

A New Battle for the Mind: Can Critical Thinking Be Taught in the Age of AI?

Rebecca A. Huggins

When I graduated from a small, private liberal arts college in the Appalachian foothills two decades ago, there was already an unsettling whisper that majoring in the humanities was a career-killer. I was twenty-one, idealistic, and working as a radio personality on the local Top-40 station. I assumed the world would unfold for me as it had for the novelistic protagonists I’d devoured in college. I had read, at the time, that the general purpose of the humanities was to keep the sciences “in check:” to stand like strange sentinels against a mechanized future. That image, once almost comic from the vantage point of my impoverished lifestyle, now reads like prophecy. STEM has long been cast as the engine of economic recovery (Bulaitis, 2020), while the humanities are still too often dismissed as a “financial dead end” (Watson, 2026). With my humble language and letters degree, I did what many humanities majors do: I entered the world of education. 

Twenty years on, the humanities are enjoying a strange revival, driven, in part, by the anxiety around generative AI, and by a broader sense that, as a nation, we are growing uncomfortably complacent. Here, a familiar argument has resurfaced: a humanities background enables us literary folk the keen skill of scrutiny, as if critical thinking were the exclusive domain of literary study. Problem-solving, on the other hand, is still typecasted to STEM, as if inference and practical reasoning were strangers to the arts. The revival feels less like vindication and more like a last‑minute plea to remember what we have almost forgotten.

Yet the belief that problem-solving belongs exclusively to STEM, while critical thinking and reasoning are the sole province of the humanities, is a persistent misconception that empirical research largely contradicts (Abrami et al., 2015; Frey et al., 2022). In reality, these cognitive functions are not exclusive territories but are deeply intertwined across disciplines. Each field applies its own rules of evidence and methods of inquiry to the same underlying work of reasoning (Erdem, 2024; National Academics, 2012; McGrath, 1959). Although problem-solving is often associated with algorithmic or quantitative tasks in STEM, every discipline requires problem-solving because each rest on foundational concepts that structure its discourse (Frey et al., 2022). Let us not forget that, historically, the original seven liberal arts included astronomy, geometry and arithmetic alongside grammar and rhetoric (McGrath, 1959). Ultimately, the humanities and sciences share common aims: the search for truth and the cultivation of a reasoned life (McGrath, 1959).  

The nation’s fixation on problem-solving and critical thinking is perhaps most acute because of three, current intersecting pressures: the disruptive rise of Generative AI, a widening global skills gap, and a growing concern about the health of democratic institutions. 

In the last few years, AI’s pervasiveness has sparked significant anxiety on college campuses: 90% of professors worry that it will weaken students’ critical-thinking skills (Wildavsky, 2026). A growing body of research suggests cognitive offloading, or letting technology perform mental tasks for us, is fundamentally altering how we process information (Guselli, 2026). Experts agree that AI can be a powerful tool when used correctly, yet empirical studies also show users are increasingly “surrendering” their thinking processes to these systems (Guselli, 2026).  AI has already begun to displace narrow technical skills such as basic coding, evidenced by a 27% drop in computer programmer employment between 2022 and 2024 (Wildavsky, 2026). As machines deliver technical outputs cheaply, the human advantage shifts toward durable skills: judgment, empathy, and the ability to scrutinize AI outputs for rigor and quality (Watson, 2026). But is that the future we want? Using AI to think for us, so we no longer must be interrogated by a reasoned argument? Worse still, will we all be tethered to large, permanent screens, reduced to the unsavory work of policing AI for accuracy? 

But I digress. 

Bolstering this fear is a concern that traditional educational models are producing graduates who are “test-prep machines” but are unable to solve real-world problems (Price et al., 2021). International indicators like PISA and national assessments like the NAEP show stagnant performance on tasks requiring higher cognitive demands (Peltier & Vannest, 2017; Jitendra et al., 2012). Beyond economics, scholars warn of a crisis in education’s role in sustaining a healthy democracy (Bulaitis, 2020). Without critical thinking, societies risk perpetuating prejudice and making irrational decisions based on misinformation or “bogus claims” from cultural and political leaders (Pedraja-Rejas et al., 2025). Philosophers like Martha Nussbaum argue that the humanities foster a deeper understanding of human concerns, enabling citizens to rise above “parochial” perspectives (Bulaitis, 2020). Alongside calls to restore a classical edge to schooling, others urge reclaiming the classroom as a laboratory for self-governance, equipping citizens to participate in ongoing conversations about freedom and equality (Abrams & Montas, 2023). 

Ultimately, the new economic order of this century necessitates workers who can tackle novel obstacles while carrying out complex communication—even in the face of an apparently fatalistic AI future. Employers now rank analytical thinking above big data or AI for skills training (Toh et al., 2025). That raises a pressing question: can we actually teach our students the skills of problem-solving, analytical reasoning and critical thinking? Or are we destined to surrender the last vestiges of humanity to the algorithm? 

Perhaps the most pervasive misconception about problem-solving and critical thinking is that these are free-floating or general skills that can be taught in a vacuum (National Academies, 2012). In truth, thinking is the ability to apply logical inquiry, which is constrained by the acquisition of pertinent domain-specific knowledge (Abrami et al., 2015; National Academies, 2012). For example, the rules for constructing an argument in physics differ fundamentally from those in history (National Academies, 2012). Educators who prioritize abstract “thinking skills” over a knowledge-rich curriculum often find that skills do not transfer to new contexts because students lack the necessary mental “schemas” to process new information (Erdem, 2024; Abrami et al., 2015; National Academies, 2012). Many educational reforms have historically assumed that students will naturally “pick up” higher-order thinking skills through the course of their normal academic studies (Toh et al., 202). Modern research, however, indicates that such skills do not develop naturally; they require explicit and systematic instruction and sustained practice (Mills & Kim, 2017). The most effective approach is often a “mixed” model, where critical-thinking instruction is integrated directly into subject-specific content (Xu et al., 2023). Teaching “thinking” in isolation creates a false sense of competence: students may learn the language of a skill without the content knowledge to apply it, producing confidence with zero predictive power for actual performance (Aposika, 2026).

A proven method for bridging content and these high-value skills in K-12 classrooms is schema-based instruction (SBI). In SBI, students learn to recognize a problem’s underlying mathematical or logical structure rather than its surface features (Peltier & Vannest, 2017; Jitendra et al., 2012). Grounded in schema theory, SBI holds that experts solve problems by drawing on well-organized structures of knowledge stored in long-term memory (Jitendra et al., 2012). A high-quality SBI framework includes four interrelated instructional practices. 

Primed structure: Teachers explicitly help students distinguish relevant semantic structure from irrelevant surface details. In STEM this might mean prompting students to identify a problem type (i.e., a ratio or percent-of-change schema); in humanities, it could mean helping students recognize recurring rhetorical or historical structures. 

Visual representations: Students use schematic diagrams or graphic organizers to map relationships among quantities or elements of an argument.

Explicit heuristics: General prompts like “understand, plan, solve” often fail because they lack actionable detail; effective instruction provides targeted, step-by-step strategies and models that embody disciplinary knowledge. In STEM, this might look like teacher modeling of a problem type or predictive models; in the humanities, it looks like modeling standards for historical thinking or the logical steps of the Socratic method.  

Procedural flexibility: Deep knowledge supports comprehension, flexibility, and critical judgement. Students must learn not only how to solve a problem, but which method is best. In STEM, students may compare and reflect on methods to build adaptable problem-solving skills; in the humanities, they practice debating competing interpretations of a text.

Together, these practices help students build adaptable reasoning skills that transfer to novel situations and promote accurate problem-solving across domains. Yet crucially, their success hinges on sufficient knowledge building and connection to existing knowledge structures.

While modern arguments for these skills are often framed in economic terms, there is also a simpler, deeply human impulse to preserve our capacity for independent thought. Classical educators, for instance, contend that critical thinking, reasoning and problem-solving serve not only vocational ends but the cultivation of intellectual and moral virtues such as courage, humility and self-control (Hess, 2024). There is also growing evidence that these capacities are being eroded: students using AI tools experience a 55% drop in brain activity compared with students working independently (Guselli, 2026). Scientists and humanists alike warn against “cognitive surrender,” the tendency to accept machine-generated answers without scrutiny (Hess, 2024).  

Are we already here — standing at the lip of the precipice that George Orwell and Ray Bradbury sketched for us? In the hush of screens and algorithmic certainty, the old landmarks of attention and slow reading feel like relics. As Bradbury warned, “You don’t have to burn books to destroy a culture. Just get people to stop reading them.” Perhaps the knowledge of real books has not vanished so much as been dispersed into a thousand shallow currents; perhaps we cling to an ideal that was always more fragile than we believed. Still, there seems to be a stubborn human ache for depth; for pages that resist skimming, for sentences that demand time, and that ache may be the last thing to go. Only time will tell.

References

Abrami, P. C., Bernard, R. M., Borokhovski, E., Waddington, D. I., Wade, C. A., & Persson, T. (2015). Strategies for teaching students to think critically: A meta-analysis. Review of Educational Research, 85(2), 275–314. https://doi.org/10.3102/0034654314551063

Abrams, A., & Montás, R. (2023, March). The defenders of liberal education are destroying it. The Atlantic.https://www.theatlantic.com/ideas/archive/2023/03/liberal-education-desantis-humanities-western-canon/673395/

Aposika, F. (2026). Investigating the relationship between problem-solving skills and academic performance in STEM subjects among secondary school students. Discover Education. https://doi.org/10.1007/s44217-026-01487-w

Bulaitis, Z. H. (2020). Value and the humanities. Palgrave Macmillan. https://doi.org/10.1007/978-3-030-37892-9_2

Erdem, C. (2024). A comparative meta-analysis of the effects of problem-based learning model on K-12 students’ cognitive outputs. Educational Studies, 50(6), 1498–1519. https://doi.org/10.1080/03055698.2022.2103650

Frey, R. F., Lemons, P. P., et al. (2022). Teaching problem solving. CBE—Life Sciences Education, 21(fe1).https://doi.org/10.1187/cbe.22-02-0030

Guselli, D. (2026, April 21). Is AI making us dumber? New study finds shocking brain impact. EastCoastRadio.  https://www.ecr.co.za/shows/danny-tee/is-ai-making-us-dumber-new-study/

Hess, R., & Jackson, R. (2024, May). Classical education is taking off. What’s the appeal? (Opinion). Education Week. https://www.edweek.org/teaching-learning/opinion-classical-education-is-taking-off-whats-the-appeal/2024/05

Jitendra, A. K., Star, J. R., Dupuis, D. N., & Rodriguez, M. C. (2012). Effectiveness of schema-based instruction for improving seventh-grade students’ proportional reasoning: A randomized experiment. Journal for Research on Educational Effectiveness. https://doi.org/10.1080/19345747.2012.725804

McGrath, E. J. (1959). The crucial importance of the humanities in a science-dominated world. The Modern Language Journal, 43(4), 162–166. https://doi.org/10.1111/j.1540-4781.1959.tb04369.x

Mills, K., & Kim, H. (2017, October 31). Teaching problem solving: Let students get ‘stuck’ and ‘unstuck’. Brookings Institution. https://www.brookings.edu/articles/teaching-problem-solving-let-students-get-stuck-and-unstuck/

National Academies of Sciences, Engineering, and Medicine. (2026, April 21). Transferable knowledge and skills key to success in education and work. https://www.nationalacademies.org/news/transferable-knowledge-and-skills-key-to-success-in-education-and-work-report-calls-for-efforts-to-incorporate-deeper-learning-into-curriculum

Pedraja-Rejas, L., Maulén, C., & Rivas, C. (2025). Critical thinking in initial teacher training: An empirical study from Chile. Behavioral Sciences, 15(5), 603. https://doi.org/10.3390/bs15050603

Peltier, C., & Vannest, K. J. (2017). A meta-analysis of schema instruction on the problem-solving performance of elementary school students. Review of Educational Research, 87(5), 899–920. https://doi.org/10.3102/0034654317720163

Price, A. M., Kim, C. J., Burkholder, E. W., Fritz, A. V., & Wieman, C. E. (2021). A detailed characterization of the expert problem-solving process in science and engineering: Guidance for teaching and assessment. CBE Life Sciences Education, 20(3), ar43. https://doi.org/10.1187/cbe.20-12-0276

Toh, S. S. S., Salden, R. J. C. M., Lansdown, T. C., Lee, C. P., & Hall, D. A. (2025). Conceptualizing critical thinking skills: An empirical study of Malaysian undergraduate students and academic staff. Higher Education for the Future, 13(1), 74–91. https://doi.org/10.1177/23476311251379682

Watson, E. (2026). The value of critical thinking: Humanities in a technological era. Washington Monthly.https://washingtonmonthly.com/2026/02/24/the-humanities-in-a-technological-era/

Wildavsky, B. (2026, April 15). What AI can’t do. Washington Monthly. https://washingtonmonthly.com/2026/04/15/liberal-arts-ai-critical-thinking-skills/

Xu, E., & Wang, W. (2023). A meta-analysis of the influence of collaborative problem solving on students’ critical thinking. Humanities & Social Sciences Communications. https://doi.org/10.1057/s41599-023-01508-1