Is AI Quietly Making Us Dumber?

You have probably seen the claim by now, usually attached to a grim headline: using ChatGPT makes your brain worse. It traces back to a real piece of research from the MIT Media Lab, a study with the memorable title "Your Brain on ChatGPT," and the fear it set off has outrun what the study actually showed. So it is worth slowing down and looking at what the researchers did and did not find.

The setup was straightforward. They asked 54 people to write short essays, split into three groups. One group used ChatGPT, one used a traditional search engine, and one worked with no tools at all, the "brain-only" group. While the subjects wrote, the researchers recorded their brain activity with EEG. Over several sessions the pattern was consistent. The brain-only writers showed the strongest and most connected neural activity, the search-engine group sat in the middle, and the ChatGPT group showed the least engagement. By the final session, many in the AI group had largely stopped composing at all and were pasting the model's text in wholesale.

That sounds alarming, and the authors did not shy from it. They described a kind of "cognitive debt" building up when people lean on the tool, and worried aloud about what happens if children meet AI before they have learned to think without it. I take that concern seriously. But notice what the study measured: brain activity during a specific essay-writing task, over a few weeks, in a few dozen people. That is not the same as a measurement of general intelligence, and the researchers did not claim it was.

What the study cannot tell you

The sample was small. The work was released as a preprint, before the scrutiny of peer review. And "lower neural engagement while a machine does the work" is roughly what you would expect to see, in the same way that your legs work less hard on an escalator. The open question, the one the headlines skip, is whether that lighter load leaves any lasting mark once you step off. This study was not built to answer it.

There is also history to lean on. Every tool that offloads mental effort has drawn the same charge. Plato has Socrates fretting that writing itself would ruin memory. Calculators were going to end arithmetic. Satnav was going to erase our sense of direction, and there the evidence is genuinely mixed, which is a useful reminder that the honest answer here is "we do not know yet," not "all fine" and not "catastrophe."

What the research does support is narrower and more practical. If you hand the whole task to the model and switch off, you engage less and you probably learn less from that sitting. That is a real finding about how to use the tool, and it lines up with other careful work we have covered, from how AI summaries can reshape what you remember to the stubborn difficulty of showing that these tools make people measurably more productive.

So is AI making us dumber? On the evidence so far, that is too big a claim for too small a study to carry. What it can fairly tell us is quieter and more useful. A tool that will do your thinking for you will, if you let it, and the cure is not to avoid it but to keep doing some of the work yourself. The same advice, roughly, that people have been giving about every shortcut ever invented.