MIT study warns of 'cognitive debt cycle' from student AI use

A startling 2025 study from the MIT Media Lab has sounded the alarm on student AI use. It found students writing essays with ChatGPT showed the lowest brain engagement across 32 neural regions.

Researchers noted that with each session, student writing became increasingly formulaic, as the core cognitive processes of learning were being completely bypassed by the technology. The lead researcher's conclusion was stark: "The task was being executed, but nothing was being integrated into the brain's memory networks," they explained to the outlet.

Another recent study from Oregon State University put a name to the mechanism, dubbing it the 'cognitive debt cycle', a downward spiral of dependency. Researchers warned that routine use of AI tools progressively weakens a student's habits of reflection and questioning, which in turn drives an even greater dependence on AI.

The warnings come as a RAND Corporation survey showed 62% of students used AI for homework by December 2025, a sharp rise from just 48% in May of the same year. Despite the high usage, 67% of students believe AI harms their critical thinking skills, and a College Board report found a staggering 84% of college faculty agree with them.

Educators first noticed the issue with writing assignments, according to a report in the Hindustan Times. They quickly realised the danger was far deeper than simple cheating. An expert in the report warned of the risk, stating plainly, "Outsource the writing, and you outsource the thinking that produces it."

Before a solution could be found, a second problem appeared: students had reportedly stopped reading, instead using AI to summarise texts in just a few bullet points.

This shortcut short-circuits another key process of intellectual development, as reading builds connections across ideas in ways a simple digest cannot hope to replicate.

Rather than banning the tech, some institutions are creating new policies. The State University of New York (SUNY) is even embedding AI literacy into its curriculum, and some universities now require students to document how and why they used AI. This forces them to engage purposefully rather than just completing a task quickly.

Research shows when AI is used as a 'thinking aid' instead of a substitute, higher-order thinking is enhanced, a distinction said to be absolutely critical.

In response to this, 31 US states have published AI guidance, and the American Federation of Teachers has even partnered with OpenAI on a £18 million training initiative for educators.