How to Use AI Without Literally Losing Your Mind
The Day My Brain Took a Nap and Refused to Wake Up
Something is being taken from you right now.
Not your job. Not your data. Not your privacy.
Something far more valuable than any of those things.
The silent thief doesn’t announce itself.
It just sits there — patient, helpful, always available — quietly doing the one thing you should never let anything else do for you.
Your thinking.
And the terrifying part?
Most people are not just allowing it. They’re grateful for it.
Not a yet a member? View the entire article for free Click here.
Picture this.
A university professor walks into a lecture hall, asks her students a question about the French Revolution, and watches as thirty hands dive — not into their memories, not into their notes — but into their laptops. Fingers flying.
Not writing.
Typing the question directly into ChatGPT.
Every student. Same tool. Same response. Same answer.
She told a reporter afterwards: “Everyone kind of sounds the same now.”
That sentence has been rattling around in my head ever since.
Here’s what nobody’s saying out loud in the middle of all this AI-is-destroying-education panic.
This isn’t a story about students. It never was.
It’s a story about you.
About me.
About what quietly happens to a human brain when it stops being asked to do the hard work.
MIT researchers strapped EEG headsets onto 54 people and watched their brains in real time as they wrote essays — one group using ChatGPT, one using Google, one using nothing but their own grey matter.
The results weren’t subtle.
The ChatGPT group showed the weakest brain activity across every region associated with memory, attention, critical thinking, and creativity. Weakest. By a distance.
And here’s the detail that made me put my coffee down: 83% of the AI users couldn’t quote a single line from the essay they had just finished writing.
Not a word. Their own essay. Total blank.
Their brains had gone into screensaver mode the moment they hit enter — and here’s the part that should give everyone pause — when researchers later asked those same people to write without AI, their brain activity didn’t bounce back.
The cognitive debt stuck around.
The brain had essentially decided: oh, we’re not doing the thinking thing anymore?
Cool. Taking a nap.
And then it just… didn’t get up.
Now before anyone gets their pitchfork out — this is a small, preliminary study.
54 people. Not yet peer-reviewed. Scientists are already debating the methodology.
Fair enough.
But here’s the thing about that.
Every other study pointing in the same direction. A separate survey of 319 working professionals found the more they trusted AI to handle tasks, the less critical thinking they applied — consistently, measurably, across 936 real-world examples.
A study of software developers found that those who fully handed their coding work to AI produced functioning code — but scored 17% worse when tested on whether they actually understood what they’d built.
Output without understanding. The answer without the knowledge.
And these were experienced programmers. Adults with years of expertise to fall back on. They still lost ground.
Imagine encountering a subject for the very first time, with no existing knowledge to measure the AI’s output against.
That’s a very different situation. That’s a lot of children right now.
A teacher in Minneapolis — fed up, sceptical, probably slightly sleep-deprived like every teacher on the planet — decided to run an experiment.
She pulled every device from her AP Literature class. Phones gone. Laptops gone. Pencils and paper only.
In September, 46% of her students felt confident about their reading ability.
By February?
Ninety-five percent.
Ninety. Five.
The students pushed back hard at first, obviously.
Teenagers deprived of their devices have the general energy of a cat being put in a carrier.
But within weeks, something shifted. They started describing — and this is their word, not mine — falling in love with reading again.
Removing the tool rebuilt the muscle.
Not because technology is evil. But because the muscle had simply never been asked to work.
Here’s where it gets genuinely uncomfortable.
There’s a psychology finding that’s been sitting in the research since 2017, mostly ignored, that I think reframes this entire conversation.
Get Simon P’s stories in your inbox
Join Medium for free to get updates from this writer.
Researchers found that simply having a smartphone on the desk — face down, powered off, completely dark — measurably reduced the cognitive capacity of the person sitting next to it.
The phone didn’t do anything.
It just existed nearby. And that was enough.
If a switched-off device can quietly drain mental resources, what does an always-on, always-available AI assistant do to a mind that is still in the process of learning how to think?
That’s not a rhetorical question.
We genuinely don’t know yet. And that’s the problem.
A RAND Corporation survey from early 2026 found that 70% of students are worried AI is eroding their own critical thinking skills.
Not researchers. Not parents. Not panicking columnists.
The students themselves.
And that number jumped from 48% to 65% among high schoolers in a single year.
When the people experiencing something are raising their hands and saying we think this is a problem — that’s not moral panic. That’s data.
Look — nobody should be shamed for using AI to lighten a mental load that was already crushing them before the technology existed. The world is demanding, exhausting, and relentless. Reaching for a tool that gives you a cognitive break is not a moral failure. It’s profoundly human.
But there’s a crucial difference between using AI as a tool and using it as a replacement for thought.
And a 2024 study actually proved this matters.
Students who critically engaged with AI — who questioned it, argued with it, edited its outputs, pushed back on its conclusions — actually performed better and felt less mentally drained.
Not the students who copied. The students who wrestled.
The tool isn’t the enemy. The process is everything.
So what does this all mean in practical terms?
The honest answer is that we’re using the same general-purpose chatbot for everything — professional research, casual questions, children’s homework, clinical decisions, university essays, emotional support — with zero differentiation between a post-doctoral physicist and a twelve-year-old working through their first algebra problem.
One tool.
Every context.
No guardrails designed for human development.
What we actually need are tools built differently from the ground up.
Imagine an educational AI that, when a student asks “what caused the French Revolution?”, doesn’t hand them a five-paragraph summary.
Instead it asks: what do you already know about France before 1789? What was happening economically? What were ordinary people experiencing?
It builds on what the student offers. Corrects mistakes gently. Fills in gaps. But always — always — makes the student do the thinking.
That’s a completely different product from “here’s your answer, want it in bullet points?”
Tools like this exist in early forms.
They’re just not where the money and attention is going. The investment floods toward making models faster, cheaper, more capable.
The balance needs to shift.
We’ve seen this pattern before.
Social media arrived with genuine potential — connecting people, amplifying voices, building communities.
Deployed without guardrails, optimised purely for engagement.
And it took a decade of documented harm to children before the first meaningful regulation arrived.
The UK just banned social media for under-16s.
Twenty US states now restrict phones during school hours.
France, Spain, Greece, Australia — all moving in the same direction.
A decade of evidence. Then action.
With AI, we’re earlier on the same curve. But this time we can see the pattern forming.
The question — the only question that really matters right now — is whether we act on what we’re seeing, or whether we wait for the damage to stack up high enough that the headlines get even harder to ignore.
The technology is extraordinary. Genuinely awe-inspiring. The potential is real.
But potential without intention is just capability waiting to be misused.
We have a choice right now that we didn’t have with social media. We can see what’s coming.
The question is what we do about it while there’s still time to build it differently.
Start here: next time you reach for AI on something that matters, write your first draft yourself first.
Even rough. Even messy. Then use AI to refine it.
That single habit is the difference between a tool that sharpens you and one that slowly replaces you.
FAQ
Does using AI always harm your thinking?
No.
The research is clear that how you use it determines the outcome.
Passive acceptance of AI outputs correlates with cognitive decline.
Active engagement — questioning, editing, pushing back — is associated with better results and less mental fatigue.
Is the cognitive damage from heavy AI use permanent?
The MIT study suggests brain activity doesn’t bounce back immediately after AI is removed — but this is early data.
Cognitive capacity follows a “use it or lose it” principle, and deliberately practising independent thinking rebuilds it over time.
Should children be kept away from AI entirely?
Blanket bans probably aren’t the answer, but the evidence strongly suggests that unrestricted access during critical developmental years carries real risk.
Children may never build reasoning pathways that AI prevents them from forming — that’s a fundamentally different situation from adults who already have those foundations.
Are there AI tools that actually make you think better?
Yes.
Tools built around the Socratic method — designed to ask questions rather than provide answers — show real promise.
Khan Academy’s Khanmigo is an early example.
They’re just vastly outnumbered by general-purpose chatbots.
Why aren’t purpose-built educational AI tools more common?
Incentives.
General-purpose models attract the bulk of investment because the commercial market is enormous.
Specialised educational tools serve a narrower audience and generate less excitement for venture capital.
That’s exactly what needs to change.
