Is AI Making Me Dumb? What the Research Actually Shows About Cognitive Cost

I noticed it about three months in. I'd be drafting an email — something I used to fire off in two minutes — and I'd find my cursor hovering over the ChatGPT tab. Not because the task was hard. Because my brain had quietly outsourced the first draft and didn't want it back.

If you've felt the same creeping fuzziness — a sense that your own thinking has gone a bit soft since you started leaning on AI — you're not imagining it, and you're not alone. A wave of new research and commentary is starting to put numbers and names to what a lot of us suspected: convenience has a cognitive price tag.

The 10-minute experiment

The most striking recent data point comes from a study covered by WIRED, which suggested that using AI for as little as 10 minutes might make you measurably lazier and less sharp on subsequent tasks. The framing is provocative, but the underlying mechanism isn't surprising to anyone who has studied cognition. When you hand off effort, you also hand off practice. And practice is what keeps the mental machinery oiled.

This isn't about AI being evil. It's about what psychologists call cognitive offloading — the very ordinary human habit of letting tools do our thinking for us. We did it with calculators. We did it with GPS (remember when you used to know how to get to your mate's place without Google Maps barking at you?). The difference with generative AI is the scope. A calculator offloads arithmetic. ChatGPT offloads framing, structuring, reasoning and writing — the load-bearing parts of thought.

Why writers and thinkers are nervous

Computer scientist and author Cal Newport, interviewed by The Chronicle of Higher Education, has been one of the more measured voices raising the alarm. Newport isn't a luddite — he writes about deep work and digital minimalism precisely because he believes attention is a finite, valuable resource. His worry, as the headline of that interview bluntly puts it, is whether AI is making us stupid.

The concern Newport and others raise isn't that AI produces bad output. Often it produces very good output. The concern is what happens to us while the AI produces it. Writing isn't just communication; it's the process by which you discover what you actually think. Skip the writing, and you skip the thinking. Sleek Magazine made a similar argument in a piece bluntly titled "ChatGPT Is Making Us Stupid" — that the act of wrestling with a blank page is the act of becoming smarter, and we're outsourcing precisely that wrestle.

Even The Wall Street Journal, hardly a publication prone to tech panic, ran an opinion piece asking the same uncomfortable question: is AI making us dumb? When the business pages start worrying, it's worth paying attention.

What's actually happening in your brain

You don't need a neuroscience degree to understand the mechanism. Three things appear to be in play:

The WIRED piece is particularly worth sitting with because the timeframe is so short. Ten minutes. That's not chronic dependency — that's a single session. The implication is that AI doesn't slowly degrade your thinking over years; it changes how you approach the very next problem.

The Australian context: it's already in the classroom and the office

For Australian readers, this lands particularly hard because adoption here has been swift. Universities from Sydney to Perth have been wrestling with AI policies. Workplaces are quietly running on Copilot and ChatGPT. The Productivity Commission has flagged generative AI as a major productivity lever — which it may well be — but productivity gains and cognitive gains aren't the same thing. You can produce more while understanding less.

The risk is generational. A student who has never sat with the discomfort of a half-finished essay at 11pm doesn't just miss the essay. They miss the cognitive workout that essay was secretly delivering. The same is true of a junior lawyer who never drafts a clause from scratch, or a graduate marketer who never has to stare down a blank brief.

This isn't an argument for going back

Here's where I want to be careful. The answer isn't to delete your AI tools and write everything longhand by candlelight. That's the Newport caricature, and it's not what he or most thoughtful critics actually argue. AI is genuinely useful. It accelerates research, breaks creative blocks, translates jargon, and saves hours on tasks that were never going to make you smarter anyway (looking at you, expense reports).

The answer is to be deliberate about when you offload and when you don't. A useful test: is this a task where the output matters, or is this a task where the doing matters?

Building a cognitive gym

If the research holds up — and the early signals from WIRED, the WSJ commentary, and the worries Newport articulates suggest it will — then we're going to need new habits, the way we eventually needed habits around sugar, smartphones and social media.

A few that have worked for me, and that map onto what the critics are recommending:

The honest conclusion

I'm not going to stop using AI. Neither, probably, are you. But the conversation has shifted in an important way over the past year. We've moved past "will it take our jobs?" — a debate still very much unsettled — to a quieter, more personal question: what is it doing to the person using it every day?

The early research suggests the answer is: more than we thought, and faster than we thought. Ten minutes, apparently, is enough to nudge you toward laziness on the next task. That's not a reason to panic. It's a reason to be a bit more like an athlete about your own mind — to notice which exercises you're skipping, and to put them back in the routine before the muscle is gone.

The tool is extraordinary. Just don't let it do the one thing you actually came here to do, which is think.

Related on Bleen

Sources