Critical thinking, the asset AI cannot replace, is the first casualty of its use in classrooms

By Education Week · July 8, 2026.

The data are telling and come from a source hardly given to alarmism: the RAND Corporation. In February 2025, 48% of U.S. middle school students said they were worried that AI use was harming their critical thinking. Ten months later, in December 2025, that figure had risen 20 points, to 68%. Among high school students, the concern went from 55% to 65% over the same period. These are not adults alarmed by a technology they don't understand: they are the users themselves, teenagers, saying they notice how outsourcing their thinking to a chatbot is taking a toll on them. Add to that a 2025 College Board report in which 87% of principals say AI could reduce the development of their students' critical thinking, and 82% fear it will hamper deep engagement with academic content.

The institutional response is lagging, but it is picking up speed: according to Media Literacy Now, at least half of U.S. states already have media literacy laws, and eleven of them were passed just since January 2024. It is a sign that the education system has understood that classic media literacy —teaching how to tell a reliable source from a hoax— has fallen short in the face of AI-generated content, which has no identifiable author, is produced at almost zero cost, and can sound authoritative without being so. On the ground, the response from the teachers featured in the article is pragmatic and fairly sensible: in Passaic (New Jersey), an acceptable-AI-use rubric is used as a starting point for classroom conversation; in Washington, an English teacher poses a simple but effective question to her students —would you ask this of your teacher?— to draw the line between asking for help to brainstorm (yes) and asking to have your entire essay written (no). And for the problem of sources, the recommendation is to begin and end any use of AI with human judgment: human prompt, human reflection, and cross-checking what the AI answers against other sources.

Our reading is that this story, though it talks about classrooms, is really a story about the future of work, and it connects directly with something we have been pointing out in our analysis of AI's impact by sector: what resists automation is precisely judgment, expert reasoning, the ability to evaluate ambiguous information and make a reasoned decision. That is the asset that will be worth the most in an economy where AI handles the routine. The problem is the paradox this article reveals: if today's students outsource critical thinking to the very tool that will demand that critical thinking of them tomorrow, they are eroding exactly the skill that should protect them. It is not a problem with the technology itself, but with how it is integrated into pedagogy: used as a substitute for cognitive effort, it atrophies; used as scaffolding —to explore ideas, question answers, compare sources— it can reinforce precisely that capacity for independent reasoning.

In the short term, the diagnosis is uncomfortable and must be said plainly: an entire generation of students is growing up with access to a tool that can think for them, and neither the curricula nor teacher training have had time to adapt to the pace at which the technology changes. Teenagers themselves notice it and put it into words, which is an early warning sign worth taking seriously, not dismissing as a passing moral panic. But in the long term, if schools manage to reshape the relationship with AI —from crutch to cognitive scaffolding tool— the result could be a generation better trained to live in a world of information and resource abundance, where scarcity is no longer of data or answers, but of the human capacity to discern which ones matter and why. That, at bottom, is the true curriculum of the coming century: not learning to compete with the machine, but learning to interrogate it better than anyone.