Schools Never Taught Critical Thinking: AI Exposed the Lie

AI Exposed the Lie: Schools Never Taught Critical Thinking

Nearly seven in ten middle and high school students now say they believe artificial intelligence is eroding their critical thinking skills. They reported this in a December 2025 survey conducted by the RAND Corporation's American Youth Panel. They also reported, in the very same survey, that they are using AI for homework more than ever before, with usage climbing from 48 per cent to 62 per cent in barely seven months. The students, in other words, can see the problem clearly. They simply cannot stop participating in it.

This is an extraordinarily revealing paradox, and it deserves more scrutiny than the predictable hand-wringing it has generated. Because the most uncomfortable question here is not whether ChatGPT is making teenagers worse at thinking. It is whether the education system that ushered AI into classrooms with such breathless enthusiasm ever genuinely valued the kind of independent, rigorous, critical thought it now claims to be losing.

The answer, if you follow the evidence, is not encouraging.

The Paradox in the Numbers

The RAND data is striking in its internal contradictions. Among the 1,214 young people surveyed (aged 12 to 29, all enrolled in school during the 2025-26 academic year), 67 per cent endorsed the statement that “the more students use AI for their schoolwork, the more it will harm their critical thinking skills.” That figure had risen more than ten percentage points in just ten months. The concern was especially pronounced among female students, 75 per cent of whom agreed, compared with 59 per cent of male students.

Yet during the same period, the percentage of middle schoolers using AI for homework leapt from 30 per cent to 46 per cent, and among high schoolers it jumped from 49 per cent to 60 per cent. Most of these students (60 per cent) also expressed concern about using AI for school-related purposes. So they are worried and they are doing it anyway. This is not cognitive dissonance in any simple sense. It is something more structurally interesting: students have correctly diagnosed a systemic problem, but they exist within a system that gives them no rational incentive to behave differently.

Consider the logic from a student's perspective. Assignments are graded. Grades determine university admissions. University admissions determine (or are perceived to determine) life outcomes. If your peers are using AI and getting better grades, opting out is not a principled stand. It is a competitive disadvantage. The students are not confused. They are trapped.

Think of it another way. You are sixteen. You have five GCSEs to revise for, a personal statement to write, and a part-time job. Your classmates are producing polished coursework in half the time it takes you to write a first draft because they are running their ideas through ChatGPT. Your teachers, overwhelmed and under-resourced, cannot reliably tell the difference. The system rewards the output, not the process. In this environment, choosing not to use AI is not intellectual integrity. It is self-sabotage.

Meanwhile, faculty at the university level are sounding alarms with even greater urgency. A national survey conducted by the American Association of Colleges and Universities and Elon University's Imagining the Digital Future Centre in November 2025 found that 95 per cent of the 1,057 faculty respondents feared that generative AI would increase student overreliance on the technology. Ninety per cent said it would diminish students' critical thinking skills. Eighty-three per cent said AI would decrease student attention spans. And 78 per cent said cheating on their campuses had increased since these tools became widely available, with 57 per cent saying it had increased significantly.

The teachers see the same thing the students see. The difference is that teachers are surprised. The students are not.

A System That Never Quite Got Round to Critical Thinking

Here is where the conversation gets genuinely uncomfortable. Long before ChatGPT existed, education reformers, cognitive scientists, and classroom teachers themselves were raising the alarm about a system that was systematically undermining higher-order thinking. The culprit was not artificial intelligence. It was standardised testing.

The No Child Left Behind Act of 2001 (NCLB) represented, in the United States at least, the triumph of measurable outcomes over meaningful learning. Under its regime, schools were judged by their students' performance on standardised assessments. The consequences of poor scores were severe: funding cuts, staff dismissals, school closures. The entirely predictable result was what educators came to call “teaching to the test,” a practice in which classroom instruction was narrowed to the specific content and formats that would appear on state exams.

The effects were devastating and well-documented. Subjects not covered by standardised tests, including art, music, physical education, and social studies, were minimised or eliminated outright. Some principals eliminated recess to devote more time to test preparation. Science was replaced with additional maths drills. Social studies gave way to language arts worksheets. The phrase that captured this era most succinctly was “sit, get, spit, forget,” a cycle in which students received information passively, regurgitated it on an exam, and promptly forgot it, having never engaged with it at any depth.

The situation in the United Kingdom has followed a parallel trajectory. Successive reforms since the introduction of the National Curriculum in 1988, the expansion of league tables in the 1990s, and the intensification of Ofsted inspections have created an accountability culture that rewards measurable outcomes above all else. Teachers in England report spending enormous amounts of time on assessment preparation, data tracking, and administrative compliance, time that might otherwise be devoted to the kind of open-ended, inquiry-driven teaching that develops critical thinking. The Department for Education published expanded guidance on AI in education in June 2025, stressing that AI tools should support rather than replace subject knowledge and that students still need a strong foundation in reading, writing, and critical thinking to use these tools effectively. But guidance is one thing; structural reform is quite another.

Paulo Freire, the Brazilian educator and philosopher, would have recognised all of this instantly. In his seminal 1968 work “Pedagogy of the Oppressed,” Freire described what he called the “banking model” of education, in which teachers deposit knowledge into the passive receptacles of students' minds, and students are expected to receive, memorise, and repeat. Freire argued that this approach was fundamentally hostile to critical consciousness; the more students worked at storing deposits, the less they developed the critical thinking that would allow them to intervene in the world as transformers of that world. His alternative, critical pedagogy, was rooted in dialogue, in treating students as co-creators of knowledge rather than empty vessels to be filled.

NCLB was, in Freire's terms, the banking model with federal enforcement mechanisms. The UK's accountability framework achieved much the same outcome through different institutional channels. And while NCLB was eventually replaced by the Every Student Succeeds Act (ESSA) in 2015, which offered states greater flexibility in assessment design, the deeper cultural damage had been done. An entire generation of teachers on both sides of the Atlantic had been trained in a system that rewarded compliance over curiosity, memorisation over analysis, and standardised answers over independent thought.

So when commentators now lament that AI is destroying students' capacity for critical thinking, the honest follow-up question is: which critical thinking? When, precisely, was this golden age of