AI and the Risk of Outsourcing Student Thinking
The Simplest Explanation Is Rarely Right
In a 2024 study, education researcher Lixiang Yan and colleagues found that university students could complete an entire course’s written work using generative AI tools. [2] The finding sounds straightforward: students use AI, therefore students learn less. That conclusion is tempting because it fits a familiar story about technology eroding effort. But the simplest explanation is often the wrong one. The concern is not merely that work gets done by someone else. The concern is what happens inside the mind when the work no longer happens there at all.
Yan’s realization points to something more specific than cheating. A student who delegates every assignment to a machine may still submit perfect answers. The grades look identical. The difference is invisible on paper but visible in the brain. This is not a moral claim about laziness. It is a mechanical claim about how thinking gets stronger or weaker. A 2026 commentary in Nature warned that heavy reliance on AI tools can blunt memory and reasoning. [1]
The science of critical thinking offers a counterweight. Yan’s study found that students who used AI to generate questions rather than answers retained more of the material. [2] The key is not banning the tool but changing the task. That distinction — between outsourcing the answer and outsourcing the thinking — is where the real question lives. The simplest explanation, that AI makes students stupid, misses this entirely. The wrong explanation is not that AI is harmless. The wrong explanation is that the harm is automatic.
