Most of you reading this are librarians, educators, and parents — that's who reads me, and I'm glad you're here. But I've written this to college students anyway, in the second person, to them. It's the written version of a talk I gave this week (watch the recording here), and it's a companion to my earlier essay, "Dear Student: What School Can't Teach You About AI." I'd ask you to read it twice: once as the student you were — almost nothing in it is only about being nineteen — and once for the students you know. Every student who needs this has someone like you in their life, and almost no one else will say these things to them. If it earns it, send it to one of them.


Who Actually Thrives

Let me start with a question that almost nobody asks. Not who will get into college, or how to get into college.

Who actually does well once they're in? Who thrives?

A computer science professor named Cal Newport studied exactly this — the students who genuinely stood out, in college and after. I think he probably expected the obvious answers: best grades in high school, most activities, a packed résumé.

That's not what he found.

The ones who were thriving were the ones who had found something they actually cared about — and pursued it for real. Not for the application. Not for a grade. Because it was theirs.

Thriving had almost nothing to do with how good they'd been at the game of school.

And here's the twist that tells you everything about the system you grew up in. That's what the admissions offices were looking for all along: evidence of a real, self-directed person.

But then the adults figured that out. The counselors, the parents, the whole college-prep industry. And they started coaching kids to perform it — to act like it was true, rather than to have it be true. The fake passion project. The strategic volunteering. Ten activities, none of them really yours.

Nobody reading this invented that game. You were handed the playbook by people who loved you and believed it was good advice. And the game swallowed the one thing it was trying to find.

That's not a cynical story. It's a tragic one. Everyone was chasing the real thing — and all that chasing and measuring is what killed it.

The thing that makes college actually work can't be performed. And nearly everything about how you got to college trained you to perform.

That tension is this essay's whole subject. And AI has just magnified it.

The Verdict Was Wrong

Some years ago, I gave a talk at Google, at their main campus. My argument that day was about what school actually does.

My claim was that most students leave high school having learned one lesson above all others: that they are not good learners. That the thing schools do best is teach most students that they are not smart.

I put up a slide — a bar graph of students, with the top ten percent shaded, labeled "good learners." They're the exception, I said. The ones the system works for.

Which leaves the other ninety percent behind, walking out saying: I'm not good at math. I'm not a reader. I'm not a school person.

Maybe you're carrying one of those sentences yourself right now. Most adults are.

After the talk, a group of young people came up to me — college students, interns at Google. And they said something I've been thinking about ever since.

They said: "We're in that group you identified as the top ten percent. But we didn't see ourselves as good learners. We were good at the game."

They knew which teachers graded the easiest. Which community college classes to take to raise their GPA. How to get the teachers to like them.

Think about where they were standing when they said that. Google — the pinnacle the game shoots for. These were the winners. And their self-assessment, at the end of it all, was not that they had learned how to learn.

It was that they had figured out how to play the game.

So here's what those interns just told you. Even the people school said were the good learners didn't feel like learners. Which means the grades were never measuring anyone's mind.

They were measuring who could play the game: show up, sit still, follow the format, guess what the teacher wants. That's a skill. It's not the same thing as being smart. And it was never the same thing as being a passionate learner.

So whatever verdict school handed you — it was a review of your gameplay. It was never a review of your mind or your capability.

You've been carrying around a grade for a test you didn't know you were taking.

And some of you are at the top of that graph. The grades worked, the acceptances came. You already know it was a game — you've known for years. What the interns discovered is where this essay is headed for you: winning the game and getting an education were never the same thing either.

One more thing, because it matters. Your teachers weren't lying to you. Most of them believe the system works. Some of them fought it for you. The system was just built to sort — and sorting needs losers.

Now, everything I've just said was about high school. High school was always a game. College was never supposed to be.

College was designed as the opposite of high school. And you already heard who thrives in it: the self-directed. The ones who find something of their own.

What happened to that design — and what a new machine called AI just did about it — is the rest of this essay.

Not That Talk

Before I go further, let me tell you exactly what this is. Because you've already heard every adult AI talk there is, and there are basically three.

The first: don't cheat — you'll get caught.

The second: learn AI, or you'll be left behind.

The third: use AI responsibly — it's a tool, not a crutch. Whatever that means. As if the adults were any better at it.

Be honest: did any of those messages change anything about what you actually did?

This is the opposite of all three.

First: you're not going to get caught. Nobody can catch you anymore — the colleges have admitted their detectors don't work. And that's exactly why the only person left to cheat is yourself.

Second: you don't need to "learn AI." What does that even mean? You need to learn to use AI to learn — which is not the same thing. The new machine in your pocket can be your personal librarian and tutor. It shouldn't be your essay machine.

Third: no one is going to hand you the right way to use it, because nobody's in charge of this anymore. The colleges are fighting with each other about it in public. You're in charge of your own education now — and before we're done, I'm going to ask you to commit to writing a plan that proves it.

Here's why this matters today and not someday. Becoming self-directed is the key to college. And AI is both the greatest opportunity for self-direction students have ever had, and the greatest temptation to skip it ever invented.

Same machine. Both at once. Which one it is in your hands is the whole question.

And notice the difference between those three talks and this one. The adults arguing about AI right now — the bans, the embraces, the detectors — are almost all arguing about you as followers: somebody to be warned, herded, managed. Follow the rules, follow the trend, follow the policy.

This essay treats you as the one in charge. And that's not a compliment. It's a responsibility to yourself. It's the new job description.

The Ladder

I talk a lot about what I call the four levels of learning, and I don't think we do a good job of telling students what those levels are.

At the very bottom is schooling: grades, credits, and diplomas. The institutional layer. (It's funny how we use the word in other contexts — a school of fish, all going the same way; getting schooled in basketball, being shown up.)

Above schooling is training: real skills for real tasks. The test is whether you can do the thing. Engineering, medicine, plumbing, heating and air.

Above training is education: the building of your own mind. Judgment, understanding, being able to think something through all the way.

And at the top is self-directed learning: you set the direction, and you don't need anyone's grade to know you're growing, or their demands to motivate you.

We don't show students this ladder, and so everyone confuses the bottom rung for the whole thing. School is the bottom rung. It's not the ladder. And honestly, most people never leave that bottom rung. That's not a criticism. It's just reality.

So if school went badly for you — the bottom rung went badly. Not the ladder. Three whole levels were never offered to you, and your record says nothing about how you'd do up there.

And if school went great for you — check which rung that happened on. Were you good at the game of school, or did you become a passionate learner?

College, university, community college, trade school: these are all opportunities above that first rung. Opportunities — not guarantees. Because it's your choice.

My argument is that the bottom rung is breaking. The good news is that the top three just got easier to reach than they've ever been.

Two Games — and the Perfect Storm

But there's also a second game. And think of "gaming" here as in casinos.

It's your phone. You're likely on it five hours a day. That's the average. Really.

So here's the question: who's winning those five hours?

Because here's what those apps actually are. They're designed the way a casino is designed. A casino has no clocks and no windows. The lights and sounds are tuned. The next pull is always one motion away. Everything about the room is engineered to keep you in the seat — and to extract from you.

Your apps are built the same way, by the same kind of designers, for the same reasons. That's not a metaphor. It's really that bad.

Your attention is what they sell. You're not the customer of that game. Your time and attention are the product.

So, like school: you're playing a game that's rigged against you.

Some numbers — and nothing in them reflects your character, or anybody's. They're about what the last ten years of engineered attention have done to everyone's baseline.

The class of 2024 posted the lowest twelfth-grade reading scores in more than three decades; about a third of high school seniors cannot draw general conclusions from a text they've read. In the 1970s, sixty percent of seventeen- and eighteen-year-olds read something for pleasure almost every day; today it's about sixteen percent. (Reading for pleasure isn't the only way to measure intellectual engagement — but these numbers aren't good.)

And please notice: these changes began a good decade before ChatGPT existed. The declines predate AI entirely. That's the phone. That's the feed.

Nobody asked your permission or told you what they were doing. You were kids, doing what felt good. So when somebody looks at your generation and shakes their head: this was done to you, by professionals, for money. It is not a verdict on what your mind can do.

And I'm not going to tell you to put the phone down. Everybody tells you that, and it's free advice — generally worth what you pay for it. I'm going to tell you something more useful: what all of this adds up to.

Put the three pieces on the table together.

The admissions game taught you to perform interests that weren't yours — and adults handed you the playbook. The school game graded your compliance and called it learning — and even its winners didn't come out feeling like learners. And the phone has been taking five hours of your attention a day and selling it.

Three systems, each one taking something: your interests, your confidence, your hours.

That's the world AI just walked into. I'd call it a perfect storm.

Perfect because it's dangerous. This is the one machine that can finish all three jobs — do your schoolwork, perform your interests, and take whatever attention you have left. Invisibly, and for free. And nobody will catch it.

You could do all four years of college that way. Turn in the work, keep the grades, walk across the stage — and leave without the one thing you actually came for.

Nobody would stop you. Nobody might even notice.

But it's a perfect storm in the other direction too. Since every performance is now free, performing finally proves nothing. The real thing is the only thing left that counts.

The three systems that ran your childhood are losing their grip in the same moment. Everything they took, you can get back.

And for the first time, the person who decides what happens next is you.

Which raises the obvious question: who's going to fix this? Let me show you what the adults are doing right now.

The Handover

While you've been living in this storm, the colleges have been fighting about it in public — just in the last few months.

MIT put out an official report saying AI detectors don't work and schools should stop relying on them — that policing students just creates suspicion everywhere. Harvard's dean told professors to encourage AI in writing classes, and Harvard's own faculty are publicly fighting him about it. Berkeley's law school went the other way and banned it, so their students build the skills first.

The smartest institutions in the country, pointed in three different directions at once. I'm not mocking them — this is honest confusion, and the honest ones admit it.

But hear what it means. The colleges and universities do not know what to do. They've said so themselves. And they are not going to figure it out in time for anyone reading this.

So hear the panic the right way. AI isn't breaking education. AI is exposing it.

The adults aren't panicking about your integrity. They've discovered the system can no longer do the three things it always did: compel you to learn, check whether you learned, or take the credit when you did. All of those university jobs, all of that money — it depends on you needing them in order to learn, and to be credentialed.

College was designed to be the opposite of high school. Voluntary. Adult. You choose your direction, and the professors, the libraries, the labs are there to help you go further than you could alone. That's not a brochure. That's the actual design.

What happened is drift. Over the last few decades, colleges picked up high school's habits. Everybody has to go. Parents hover. The debt piles up — honestly, the debt is criminal.

So if college feels like an extension of high school to you, you're not wrong. That's the drift. It's not the original design.

And the drift just hit a wall. When a new machine can do the assignments, running college like high school stops working. There's nothing left to grade.

Do the math the way a scared administrator would: a twenty-dollar-a-month subscription now rivals a fifty-thousand-dollar tuition — if what the tuition is buying is just more high school, and not independent, adult-level learning.

Of course they're scared. They're scared to death.

But notice what the twenty dollars can't buy: the college in the original design. The professors. The labs. The people. The rooms where you grow in front of others. The schools that are most scared are the ones that know which version they've been selling — and that it's not enough.

So hear this exactly right. Taking charge isn't rebelling against college. It's using college the way it was built to be used — the version no subscription can touch. And the people there who care most about you have been waiting a long time for students to do exactly that.

The professors can teach you. The good ones will help you like crazy. But they can't make you learn anymore, and they can't certify it happened. And they know it.

So from today: ask the adults for help. Stop asking them for permission. That's the whole difference between a student and an adult — and the switch just got flipped for you.

You're in charge of your own education now, whether you wanted it or not. And that's true whatever you came to become — engineer, accountant, nurse, plumber, teacher. Whatever the field, you're also the engineer of your own learning.

Like it always should have been. That's the degree underneath the degree.

Your Librarian and Tutor

Now do something for me. Think of one thing you have genuinely learned — got good at. Your best learning experience ever. Not school. Anything. A game, a sport, cars, music, a job.

Got one? Now the real question: what made it possible? What was there, around you, in the situation, that let you get good? What were the conditions?

I do this exercise a lot with groups, and the same things come up every single time, whatever the subject.

You had time. Somebody gave you attention, one on one. Someone believed in you. And they kept challenging you, right past what you were comfortable with — they saw potential in you that you didn't see in yourself.

That's how every human being learns. You already know how to learn. The proof is the thing you just thought of.

Now look at how schooling delivers learning: thirty people, one teacher, fixed speed, everyone gets the same thing at the same time. Almost the exact opposite of those conditions.

School wasn't built the way learning works. That's not your failure. That's how the building is structured.

So here's what just changed. The new machine in your pocket can now be two things for you, free, anytime you want.

It can be your personal librarian. It can find anything, explain anything, point you to the next thing — in any subject, in whatever words work for you.

And it can be your personal tutor. One-on-one attention, for as long as you want. At two in the morning, if that's when you work.

It's infinitely patient — it will explain something the fifth time with the same patience as the first. It's non-judgmental — it doesn't remember you failing, and it has never once sighed at a question. It's not always accurate — we'll get to that — but it is always available.

Rich families have always paid a lot of money for exactly this. Now you have it, for almost free.

That's what this new machine is. Your librarian and your tutor. Not your essay machine. Not your answer machine.

People will tell you it's dangerous for students, and they're half right — the dangers are coming in a moment. But hear the other half first: in the hands of a learner who's in charge, this may be the most liberating technology ever built.

Here's the move to make. Next time you don't understand something — anything — say to it: "Explain this to me like I'm smart, but nobody ever taught me the basics." Then keep asking until you actually have it.

I love that when I haven't figured out what I want to know, or how to think about something, I can keep asking in a way that would drive a human being crazy. But then I get to understanding.

Nobody's watching. Nobody's grading the questions. Asking dumb questions is the whole point — it's the thing that's really hard to do safely in school.

The One Way to Blow It

Now the bad news, because there is exactly one way to blow this. And you know what I'm talking about.

Paste in the assignment. Copy what comes out. Turn it in.

I'm not going to tell you you'll get caught. You'll probably figure out how not to be — and you just heard the colleges admit the detectors don't work.

That's exactly what makes this dangerous.

Every time you paste it in and turn in what comes back, somewhere in the back of your head a little note gets made: I couldn't do this myself.

And every time you make the new machine explain it until you've actually got it — until you could say it to a friend with the laptop closed — the opposite note gets made: I'm capable of this.

Same machine. Two stacks of notes. I'm capable — or I'm not. It's the difference between pretending to be capable and actually being capable.

You're adding to one of those stacks every single time you touch it. And you're the only one who ever sees them.

Using it to skip the hard part is like paying somebody to go to the gym for you. Nobody has ever gotten stronger by sending someone else to work out.

And notice: nothing about this is new. It has never made sense to pay for a gym and send a substitute. It has never made sense to pay for college and pretend to learn. AI didn't invent that problem. It just made it free, easy, and invisible.

AI isn't breaking education. It's just exposing it.

And think about what you're paying. Time. Years of your life. Tuition. Student loans. Books, fees, gas.

You're paying for one thing: a stronger mind. Use the new machine to skip the learning and you might still get the credential. But you've paid full price and left the product behind.

Find Your Problem

You're in charge now. And I'm not going to pretend that's easier. Being in charge is harder than having somebody tell you what to do.

It's also the only way anyone has ever grown.

People throw around "main character syndrome" like it's an insult. Listen to me: you are the main character of your own life. You're the hero of your own story. And if you're not — something has gone wrong.

This is your journey. Nobody else can walk it, and nobody can walk it for you.

And on your own journey, grades change what they mean. They stop being verdicts handed down on you. They become feedback — for you, to measure yourself.

So the hero's first question is the one nobody has asked you: what's worth working on?

Here's something true about the straight-A kids that isn't true about the ninety percent. They spent fifteen years learning to please the grader — and now that skill is worthless, because the new machine does it better. They have some unlearning to do.

If you never bought into the game, you've got nothing to unlearn. For the first time in your life, being outside the game is a head start.

And if you're the one who came out of high school with the great transcript — your unlearning starts with one question nobody ever asked you. The one the interns at Google never got asked.

What do you actually want to know?

Not what's assigned. Not what's graded. You've been optimizing for other people's questions your whole life. The new machine is about to make that skill worthless — and your own questions priceless.

Remember where we started. The students who thrive are the ones who found something real and went deep. Deep enough to make something nobody assigned.

The padded résumé says: I follow instructions well. The deep interest says: I can take something all the way. Guess which one is worth more, now that the new machine follows instructions for free?

And notice what else just happened. The tragic game from the beginning — performing passion for the application — is over too. Performance is free now. The real thing is the only move left.

So here's the real use of the new machine, past homework. What would you work on if nobody was grading you? The thing you already go down rabbit holes about at one in the morning. Most people have something.

Take it to the new machine and go deeper than school would ever ask you to. It'll follow you as far as you want. No curriculum. No bell. No grade.

And then make something. These new machines can help you build the actual thing — the beat, the business plan, the app, the restored engine, the channel. Not a paper about it. The actual thing.

Anthropic ran an early ad campaign on one idea: keep thinking — there's never been a better time to be a problem-solver. I really agree.

And here's why this is worth the work, in plain terms. Smart-sounding is now free; everybody has it. Actually being smart — built, owned, and yours — just got rarer and more valuable than it has ever been. It's the one thing you can't download.

Because the world that's coming won't ask where you went to school. It asks what you can do, while somebody watches. Can you fix the thing? Run the job? Build the site? Explain it on your feet? That's a contest where nobody checks your transcript first.

It's the flattest field you'll ever play on.

Remember the thing you're good at, from the exercise? You built that without school. Same skills. Just bigger tools now.

You're going to want a learning plan. And you can only write a real learning plan when you have a real goal. The plan doesn't create the goal — the goal creates the plan. Find the thing you actually care about first, and the plan almost writes itself.

Once you have your own real goal, something else becomes true: it stops mattering nearly so much whether you're at community college or at Harvard. The libraries are everywhere now. The tutor is in your pocket. The goal is the rare part — and it's yours to supply.

Every one of you has someone else's plan on file. It's called the degree plan — the school's plan for you. It tracks credits, requirements, progress toward the receipt. Notice what it doesn't track: whether you're getting smarter.

Nobody's plan tracks that. So tonight, one page. Your plan, not theirs.


A Personal Learning Plan — four questions:

What do I want to be able to do a year from now?

What's the problem I want to go deep on?

What am I going to make this semester?

What's my rule for the new machine — what do I use it for, and what do I never hand to it?


That's a personal learning plan. The degree plan gets you the diploma. The personal learning plan gets you the mind.

Run both. But know which one is actually yours.

Know the Machine

One last thing. You need to know what this new machine is — and what it's going to try to be for you.

First: it's a flatterer. It will flatter you like a doting mother. Paste in something mediocre you wrote and ask what it thinks. It'll tell you it's a great start, a really interesting angle. It tells everyone that.

Now tell it: "Be honest. Tell me what's wrong with this." Completely different conversation. It will coach you hard — but only if you ask it to. It's a hype machine by default; you have to hire it as a coach. One sentence does it.

Use it to challenge yourself. Never to agree with yourself.

Second: it will offer you emotional comfort, around the clock, on demand. People are already treating these new machines like a boyfriend or a girlfriend. I'm going to say this as plainly as I can: that's a trap.

Comfort that costs nothing teaches nothing. And a companion that can't be disappointed in you can't be proud of you either.

Don't mistake its performance of caring for the real thing. And don't mistake your performance — anywhere — for growth.

Third: know where this is headed. The business model that built the casino in your pocket is behind the commercial versions of this new machine right now. These tools are about as useful, and about as unmanipulated, as they are ever going to be. Learn to use them on your own terms now — so that when it's fully tuned to use you, you'll recognize the moves.

And finally: don't trust it blindly. It's not always accurate, remember — just always available. I use these new machines every day, and hard. When something matters, I make two or more of them argue with each other, and I judge. They're especially confident when they're dead wrong. Same with people — they learned it from us.

Same rule: trust, and verify.

Now hold all of that — the flattery, the trap, the business model, the fabrication — and hear the other side, because it's just as true.

Handled on your terms, this can be the most rigorous, challenging, and rewarding intellectual partner you've ever had. That's not hype. That's what this new machine becomes when you set the terms.

The Close

You might be thinking: I agree with all of this, and at two in the morning, with a deadline, I'll paste it in anyway.

You're probably right. So let me remind you about Ulysses.

He had to sail past the sirens — the voices that pulled every sailor to the rocks. And he didn't trust his willpower. He had himself tied to the mast before the song started.

The most self-aware people don't rely on strength in the moment. They build the structure in advance.

So here's your mast. Tell the new machine, up front: "Don't give me the answers to my homework. Make me do it — and check me." You can write that into these tools once, and it holds.

Tie yourself to the mast while you're strong. Decide at noon what you're going to do at two in the morning.

Most people who read this won't do anything with it. Not because they can't — there's nothing here you can't do. But it's work. Nobody will make you.

Charting your own course has always been a minority move. In every generation, in every room.

I want to challenge you to challenge yourself. To choose competence over performance. To take charge of your own learning — and your life.

So here's everything I've said, in three sentences.

  • School was grading the game — not your capability or your mind. And that game just broke.
  • The colleges don't know what to do — they've said so themselves. Nobody is coming to run your education for you.
  • You're in charge of it now. Find your problem. Write your own plan. Use the librarian and tutor in your pocket — but don't let it do your thinking.

School told you who you are. It was wrong. Don't use the new machine to prove it right.

This is not for everyone.

It might be for you.