A recent report from tech journalist Sean Endicott at Windows Central spotlighted a troubling finding from MIT researchers: Frequent use of AI tools like ChatGPT in schools could stunt mental development.
Drawing from a study titled “Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task,” Endicott explained that researchers at the MIT Media Lab found participants who used large language models (LLMs) for essay writing showed significantly lower brain activity compared with those relying on traditional methods.
In the study, 54 people between 18 and 39 years old were divided into three groups: one that relied solely on their own thinking, one that could use a standard search engine, and one that could use an LLM. The LLM group completed tasks with 32% less cognitive load and reported less frustration, but when later asked to write without AI, their performance dropped.
According to Endicott, the researchers linked this to “cognitive offloading” — outsourcing mental effort to a tool — potentially weakening critical thinking over time.
“The LLM undeniably reduced the friction involved in answering participants’ questions compared to the Search Engine,” the researchers noted, “however, this convenience came at a cognitive cost, diminishing users’ inclination to critically evaluate the LLM’s output or ‘opinions.’”
The study said that this effect could magnify the “echo chamber” problem, where algorithm-driven responses subtly limit exposure to diverse perspectives.
Endicott observed that while it may seem obvious that writing an essay requires more brainpower than having a tool write it, “science is about more than hunches” — and evidence is essential before changing education policies.
While the study’s small sample size and lack of peer review mean more research is needed, the implications are significant. Endicott also drew a parallel to a separate Microsoft–Carnegie Mellon study, which likened the decline in critical thinking from AI dependence to “muscles atrophying.”
Coupled with surveys showing students often use ChatGPT to cheat, Endicott warned that both learners and workers risk becoming “less capable of performing certain tasks on their own” if they offload too much thinking to AI.