Earlier this year, education researcher Lixiang Yan realised that students could outsource all the work for his university course to artificial intelligence. Yan, who teaches at Tsinghua University in Beijing, China, was using an AI agent called Codex to set up an online course on a learning management platform. But it quickly became clear that this capable AI could complete all the assignments, answer questions and participate in discussions on its own — “without the learner’s intellectual involvement”. The experience prompted Yan to reflect that teaching students to reason and think for themselves is “more important than ever”. And he is far from alone. In a 2025 survey of more than 1000 US university lecturers conducted by the American Association of Colleges and Universities, 90% said they believed generative AI would diminish students’ critical thinking skills. Around 20% of higher education students who use AI reported that they already found it harder to solve problems without it, according to responses from more than 8000 participants in a global survey this year. “Learning happens as a consequence of cognitive processing. Cognitive processing usually takes effort, and that is precisely what AI effectively removes,” says cognitive psychologist Daniel Willingham of the University of Virginia in Charlottesville. “People are really concerned about this.”
What is critical thinking?
Ask five people what critical thinking is, and you will get five different answers. “Critical thinking is a kind of pinnacle” of human cognition, says Willingham. “It is an umbrella term for instances in which people think well.” One reason for the range of definitions is that the term is widely used in education and everyday speech, but less often by researchers who study cognition. Instead, they talk about individual higher-level processes, such as memory, reasoning, decision-making and problem-solving, which engage distinct but overlapping mechanisms in the brain.
Certain general strategies are characteristic of critical thinking, researchers say. An important one is the ability to consider alternative possibilities or perspectives other than your own, much as a scientist weighs different hypotheses. But practising critical thinking requires skills and knowledge in the relevant subject area — so being able to do it in one field does not mean a person can do it in another. Neurologists, for example, are not particularly good at diagnosing cardiac cases, one study found.
Does AI weaken this skill?
Concerns that students are intellectually underprepared are nothing new. In 1983, a report by the US National Commission on Excellence in Education warned that many 17-year-olds lacked “the higher-order intellectual skills we should expect of them”. Over recent decades, critical thinking has become a goal for many education systems, says Stéphan Vincent-Lancrin, who led a decade-long programme on critical thinking at the Organisation for Economic Co-operation and Development (OECD) in Paris. It is now considered an educational priority by UNESCO and features in school curricula in almost all OECD countries. He cites a course in which students investigate the causes of type 2 diabetes, revising their understanding as they explore different perspectives, such as the contribution of genes and the influence of the environment.
But that does not mean the goal is being achieved. “There is little reason to think that we used to do a brilliant job” of teaching critical thinking, Willingham says grimly, “and now AI has come along and destroyed everything we achieved” for today’s students. Schoolteachers and university lecturers have countless stories of students who, they say, struggle to solve problems and evaluate information independently because they outsource the work to AI. And some studies support the idea that AI can undermine skills. A 2024 study found that school pupils performed better on maths problems when they had access to generative AI such as ChatGPT, but once that access was removed, they performed worse than those who had never had it.
Overall, research does not yet clearly show what effect generative AI has on learning and critical thinking. That is because AI tools are relatively new and varied, existing studies have conflicting findings, and too few rigorous or long-term studies have been conducted, researchers say. “The tools can help, but they can also have a negative effect,” says Yan, who has reviewed the literature. “It actually goes both ways.” Some studies suggest that AI tools can help people learn if, instead of providing ready-made answers, they offer feedback or tasks that encourage students to think and solve problems. “It all comes down to how we structure the task,” says Yan. But Willingham and Yan acknowledge that people tend to seek quick answers rather than use AI in ways that require hard work. When Yan asked students to use AI designed to stimulate their own thinking, some switched it off and turned to ChatGPT. “It is a little shocking, but it is true, right?” he says.
Future research could help. In June, the Education Endowment Foundation (EEF), a non-profit research organisation in London, announced that it would spend up to £2,5 million (US$3,4 million) on rigorous studies of the effects of generative AI on learning, comprehension, memory and problem-solving. “It is good that someone is doing these studies,” says Willingham, although he still expects bleak results. “It will take time before you see this decline,” he says, “but the decline seems fairly inevitable.”
How can critical thinking be taught?
The fact that critical thinking differs across subject areas is one reason it is difficult to teach, researchers say. “It is not universal or free-floating — it needs to be taught within specific subjects,” says educational psychologist Paul Kirschner of the Open University of the Netherlands, based in Heerlen. That means giving people the specific knowledge and reasoning skills they need in that field. In history, critical thinking might involve examining records and artefacts, he explains, whereas in science it might require understanding the relationships between theory, hypothesis, method and evidence. A second challenge is that critical thinking often involves recognising reasoning strategies.
