Your Brain on GPT

I need to be honest with you about something. I use ChatGPT. I use Claude. I use Gemini. Pretty much every day at this point. And for the past few months I've had this nagging feeling that something is off. Like my brain isn't working the way it used to. I'll sit down to write something and the words just don't come the way they did before. I'll try to remember something I read last week and it's just gone. At first, I thought maybe I was just tired or stressed or whatever. But then I started seeing the research.

Last year researchers at MIT's Media Lab ran an experiment that got a lot of attention. They took 54 college students and split them into three groups. One group wrote essays using ChatGPT. Another group wrote essays using Google for research. The third group wrote essays using nothing at all, just their own brains. They did this over four months, wearing EEG headsets so the researchers could actually see what was happening in their brains while they worked.

The results were honestly kind of scary. The ChatGPT group showed the lowest brain activity of all three groups. Their neural connectivity, which is basically how different parts of your brain talk to each other, was significantly weaker than the other groups. And it got worse over time. The more they used ChatGPT, the lazier their brains got. When they tested memory, 83 percent of the ChatGPT users couldn't quote a single line from the essays they had just written. Not one line. They had literally just written these things and they couldn't remember what was in them. Because they didn't actually write them. ChatGPT did. Their brains never engaged with the material in the first place.

The researchers called this "cognitive debt." You're essentially borrowing thinking from an AI and your brain is paying the price later.

Here's the thing that really freaked me out. In the fourth session of the experiment, the researchers switched things up. The people who had been using ChatGPT had to write without it. And the people who had been writing on their own got to use ChatGPT for the first time.

The ChatGPT users couldn't bounce back. Even after they stopped using the AI, their brain activity stayed low. Their neural networks stayed sluggish. It's like their brains had gotten so used to not working that they forgot how to work. Meanwhile the brain-only people who tried ChatGPT for the first time showed the highest brain connectivity of the entire study. They were actively integrating the AI suggestions with everything they already knew. Because they had built up that cognitive muscle over the previous sessions.

So if you've been using AI heavily and you're thinking "I'll just stop when I need to," the research suggests it might not be that simple. You might have already lost something.

The effect was strongest in younger people. People aged 17 to 25 showed the highest dependence on AI tools and the lowest critical thinking scores. People over 46 showed the opposite pattern. Higher critical thinking, lower AI dependence.

The researchers think this is because older people learned to think before AI existed. They built those neural pathways and cognitive skills the hard way, through years of actually using their brains. Then when AI came along, they had something to fall back on. But younger people are growing up with AI from the start. They're outsourcing thinking before they ever learn how to do it themselves.

As one researcher put it, you can't atrophy a muscle that was never built.

Cognitive offloading: the scientific term for getting lazy

There's a term for what's happening here. It's called cognitive offloading. It's when you delegate mental tasks to external tools instead of doing them yourself. And it's not new. We've been doing it forever. Writing things down instead of memorizing them. Using calculators instead of doing math in our heads. Relying on GPS instead of learning how to navigate.

But AI is different. Those other tools offloaded specific tasks. Writing offloaded memory storage. Calculators offloaded computation. GPS offloaded spatial navigation. AI offloads reasoning itself. The actual process of thinking through problems, evaluating evidence, constructing arguments. That's what makes it so much more dangerous.

There's research showing that London taxi drivers who had to memorize the entire city's streets had measurably larger hippocampi than bus drivers who just followed fixed routes. Use it or lose it, applied to brain structure. Now imagine what happens when we stop using not just our navigation abilities but our thinking abilities altogether.

Students are already seeing the effects

Teachers are reporting something they're calling "digitally induced amnesia." Students submit assignments and then can't recall what was in them because they never committed the information to memory in the first place. The work got done but no learning happened.

Reading skills are particularly at risk. The capacity for what researchers call "cognitive patience," the ability to sustain attention on complex ideas, is being diluted by AI's ability to summarize long text. One expert noted a shift in how students talk about reading. It used to be "I don't like to read." Now it's "I can't read, it's too long."

There's also evidence that writing is becoming more homogeneous. When everyone runs their ideas through the same AI models trained on the same data, everyone starts sounding the same. The unique voice and perspective that comes from actually wrestling with your own thoughts gets flattened out into generic AI prose.

Why younger people are hit hardest

This is where it gets really concerning for my generation. The research consistently shows that younger users are more affected than older users. And the explanation makes a lot of sense when you think about it.

People over 46 who use AI are offloading tasks they already knew how to do. They built those skills over decades of practice. The neural pathways exist. They can fall back on them if needed. But people aged 17 to 25 are offloading tasks they never learned to do in the first place. The neural pathways for source evaluation, logical argument construction, and independent problem solving were never formed.

Look, I'm not going to tell you to stop using AI. I'm not going to stop using it myself. These tools are genuinely useful and they're not going away. But the research is pretty clear that how we use them matters a lot.

The biggest factor seems to be whether you're using AI to replace your thinking or to augment it. One study found that a version of ChatGPT designed to act like a tutor, giving hints instead of answers, completely eliminated the negative cognitive effects. The problem isn't AI itself. It's using AI as a shortcut instead of as a tool.

If a student uses AI to do the work for them, rather than to do the work with them, there's not going to be much learning. No learning occurs unless the brain is actively engaged in making meaning and sense of what you're trying to learn, and this is not going to occur if you just ask ChatGPT, "Give me the answer to the question that the instructor is asking."

Some practical things that seem to help based on the research. Use AI for brainstorming and structuring ideas but write the final version yourself. Read AI output, close the window, then rewrite it in your own words. Ask AI for opposing viewpoints to challenge your thinking instead of just confirming what you already believe. Do the hard parts yourself first, then use AI to refine or expand.

The key insight from the Harvard Gazette piece on this was pretty simple. If you think you're in school to produce outputs then AI helping you produce those outputs might seem fine. But if you're in school to learn, remember that the output is just a vehicle through which learning happens. When you confuse those two things, you use AI in ways that aren't conducive to learning.

The uncomfortable bottom line

I started this piece talking about that nagging feeling that something was off with my brain. And honestly writing this has made me more worried, not less. The research is consistent across multiple studies and multiple methodologies. Heavy AI use correlates with weaker critical thinking, reduced memory, lower cognitive engagement, and potentially permanent changes to how our brains work.

The effect is real. The question is what we do about it.

I don't have a perfect answer. But I think step one is just being honest about what's happening. We're in the middle of a massive experiment on human cognition and most of us signed up for it without reading the terms and conditions. The least we can do is pay attention to the results.

Your brain is a muscle. It needs exercise. And AI is making that exercise optional in ways it never was before. Whether that's a net positive or negative for humanity is going to depend on whether we're smart enough to recognize the tradeoff and act accordingly.

Which is kind of ironic when you think about it. We need to be smart enough to recognize that we're getting dumber. And the longer we wait, the harder that recognition becomes.

This article draws on published research, including the MIT Media Lab study on ChatGPT use and neural connectivity, and reporting from the Harvard Gazette on AI and learning. Readers interested in the underlying studies are encouraged to consult those sources directly.