A provocative new study out of MIT has ignited headlines claiming that ChatGPT might be harming your brain. But as is often the case with viral AI stories, the truth is far more nuanced.
On Episode 155 of The Artificial Intelligence Show, I broke down the study with Marketing AI Institute founder and CEO Paul Roetzer to determine what's worth paying attention to.
The Study at a Glance
Titled "Your Brain on ChatGPT," the study analyzed how different tools impact cognitive engagement during essay writing. Participants were split into three groups:
- One wrote essays using only their memory ("brain-only")
- One used a search engine
- One used ChatGPT (GPT-4o)
Using EEG scans and linguistic analysis, researchers found that participants who relied on ChatGPT showed weaker neural connectivity and less engagement in memory and decision-making areas of the brain. Their essays were more uniform and less original. And they had more trouble remembering or quoting from their own writing—even just minutes after completing it.
But there’s a catch.
Why the Panic Is Premature
The paper, as AI expert Ethan Mollick points out, is being badly misinterpreted in viral posts and sensational headlines. He writes on LinkedIn:
"This new working paper out of the MIT Media Lab is being massively misinterpreted as "AI hurts your brain."
It is a study of college students that finds that those who were told to write an essay with LLM help were, unsurprisingly, less engaged with the essay they wrote, and thus were less engaged when they were asked to do similar work months later. It says something important about cheating with AI (if you let it do your work you won't learn) but it doesn't tell us anything about LLM use making us dumber overall.
This misinterpretation isn't helped by the fact that this line from the abstract is very misleading: “Over four months, LLM users consistently under-performed at neural, linguistic, and behavioral levels.” But the study does not test "LLM users" over four months, it tests (9 or so!) people who had an LLM help write an essay in an experiment writing a similar essay four months later.
To be clear this isn't a defense of blindly using AI in education, they have to be used properly to be effective. We know from this well-powered randomized controlled studies that just having the AI give you answers lowers test scores.
But that doesn't mean that LLMs rot your brain."
Roetzer agrees, pointing out that you can very quickly get a sense of Mollick's point if you actually read further about the study, rather than just consuming headlines.
"It's like saying we gave calculators to a control group who didn't know how to do math, and we found that people who relied on the calculator to do math didn't actually learn math," he says.
The real takeaway? If you use AI to bypass critical thinking, you’ll think less. That’s not a revelation—it’s just common sense.
From Research to the Real World
The study seems methodologically solid and points to a real effect. But it actually strengthens the case for responsible AI use, says Roetzer. In education and business, the real imperative is to teach people how to use AI tools to accelerate—not replace—learning and thinking.
That means:
- Starting with human comprehension
- Using AI to test, refine, or expand ideas—not to generate entire outputs blindly
- Creating environments that reward cognitive engagement, not just finished deliverables
Roetzer also introduced a helpful framework to understand where there are cognitive gaps in AI outputs that human brains may need to be aware of and/or check:
- The Verification Gap: Humans need to fact-check AI outputs.
- The Thinking Gap: Humans have a limited capacity to critically evaluate AI-generated content.
- The Confidence Gap: Without truly engaging with the underlying material, discomfort arises when you present material created by AI but don’t fully understand it.
Together, these gaps explain a growing dynamic in the workplace:
As we produce more with AI, if we don't understand and address these gaps, we risk retaining and understanding less.
In Roetzer’s own experience, this dynamic plays out even in day-to-day meetings.
"I still type out everything in every meeting I go to," he says, rather than rely on an AI notetaker. "If I just have the notetaker, there's less cognitive load. But that cognitive load is actually what embed it in my memory."
The Bottom Line: Critical Thinking Still Matters
The MIT study, when read carefully, doesn’t say AI is inherently dangerous to our brains. It says lazy use of AI is. And that’s an important distinction.
So what’s the smart takeaway? Use ChatGPT. Use it often. But never outsource your brain. Because the skill that matters most in the age of AI is the same one that mattered before it: knowing how to think.
Mike Kaput
As Chief Content Officer, Mike Kaput uses content marketing, marketing strategy, and marketing technology to grow and scale traffic, leads, and revenue for Marketing AI Institute. Mike is the co-author of Marketing Artificial Intelligence: AI, Marketing and the Future of Business (Matt Holt Books, 2022). See Mike's full bio.