MIT warns of 'cognitive surrender' in classrooms due to AI

Cognitive surrender: the AI that switches off the search for answers

A committee at MIT has given a name to a phenomenon many suspected without daring to articulate: 'cognitive surrender' —in the original, Rendición cognitiva—. It is not about doing homework with a chatbot. It is the inability to seek one's own solution because it is taken for granted that the correct answer is just a click away. The effort, quite simply, ceases to make sense.

The document, available via the portal aiandeducation.mit.edu, does not stop at diagnosis. It acknowledges that AI can already solve almost any graduate work —essays, problems, demonstrations, and code— and that classical evaluation methods are obsolete. Its proposal is not to prohibit anything. It is to implement oral exams, year-long work portfolios, and more teaching assistants for grading.

What is cognitive surrender and why is it worrying?

The key lies in distinguishing between the tool that assists and the one that replaces. For decades, a student without a calculator was a slow student; one without a search engine, a student with less data. AI doesn't just save time: it removes the entire problem. When the final result is immediately available, the phase of guessing, error, and understanding —the process that builds judgment— is completely skipped.

There is a nuance not to lose. Much of the work done today with AI is based on knowledge acquired beforehand. The problem is the generation that begins learning with the solved answer from day one, without having gone through the process that teaches critical thinking.

Blame on AI or an exhausted educational system?

Here, the consensus fractures. One part of the analysis maintains that AI is the comfortable scapegoat and that the deterioration began much earlier: decades of memorization without context, prioritizing the result over the process, and an academic machinery that rewards compliance with the syllabus. From this viewpoint, AI does not create the problem; it finishes it.

Another current responds with a difficult-to-refute argument: if the educational system had been failing generally for decades, elite universities would have noticed it first, and they didn't until the chatbot arrived. The level and scale of this warning, they argue, are incomparable to old complaints.

Why a calculator is not the same as AI

The parallel repeats itself to saturation: writing, the printing press, and the calculator also caused atrophy. True, but with a difference considered capital here. A calculator prevents multiplication; it doesn't solve the problem. It was a step, not the staircase.

The thesis running through this matter is that AI removes the entire staircase. It is no longer about shortcuts, but that the final answer is available without any journey. And when the creative process —the lifelong 'inspiration,' the muses— demands a long and painful journey because that is where value resides, having the destination without the path empties it completely.

From classroom to operating room and editorial office

The exam is not the real problem. The problem is defense. In medicine, the MIR tests endurance, not talent; a professional who passes with AI but cannot interpret lab results lacks stamina in the first on-call shift. In journalism, the risk arrives differently: if everyone generates texts based on the same four instructions, the value of the news drops to zero. Writing becomes a click-bait printer, and the reader who paid for real information walks away.

Dependency as the final trap

One scenario reappears again and again: it is not that AI makes us simpler; it is that it makes us fragile. Anyone uses a mobile phone; almost no one knows how to build it. A society that delegates reasoning to complex systems it does not understand hangs by a thread. The question is no longer whether AI atrophies a skill. It is whether we still know what to do when it shuts down. And no one has answered that.