Education After AI: Knowledge Is Free, Confusion Is Expensive

Walk into almost any school or university today and you’re touring a museum (on some days, a mausoleum) of 19th-century education. People wear different clothes, there are MacBooks on the desks, someone’s majoring in Enology, but the operating system is basically the same one they had when J.R.R. Tolkien taught at Oxford.

Now put that next to the reality outside: every student carries a device in their pocket that can explain the basics of quantum mechanics to a 12-year-old.

Using the same approach to education now as we did in the 1940s, we’re trying to run iOS on an abacus.

And please …  the question isn’t “How do we integrate AI into education?” The question is much more uncomfortable: “What is the point of existing schools and universities if AI can teach almost anything to almost anyone?”

Most schools don’t want to touch that question with a 10-foot pole. They stick their heads in the sand, pretend it is still 2022, and just ban ChatGPT. Their idea of getting up to speed with AI is using AI to detect AI-generated assignments.

Knowledge Is Free

For centuries, education has been built on a simple assumption: teachers know things that students do not, and the school’s job is to facilitate the knowledge exchange from teachers to students.

That wonderful world is gone forever. Knowledge is not scarce or valuable anymore.

If you’re a teacher and you think you have value because you know the material, you’re already obsolete. A free app on the student’s phone has read more books than you, never forgets anything, and can explain the same thing five different ways without losing patience – all of it in Morgan Freeman’s voice – if that’s what the student wants. So… please stop lying to yourself.

But what’s worse is that when not just knowledge but also tutoring – i.e., the process of transferring the knowledge into a student’s head – becomes free, the old model (sit down, be quiet, and I will teach you things you didn’t know) no longer makes economic sense. In a market economy, you can’t charge money for something that’s free and abundant.

The Essay Is Dead. We Just Haven’t Buried It Yet. And it Stinks.

Let’s talk about the fundamental and universal element of any school system: the essay.

Ten years ago, asking a student to “write a 1,500-word essay on the causes of World War I” forced them to read, think, come up with their own version of the thing, and then write. The mere effort of doing it pretty much guaranteed that they retained at least some part of the processed material.

In 2025, the same assignment makes the student: type the prompt, tweak once or twice, copy, paste, exhale. The finished essay now tells you nothing about what the student actually knows and, in a depressingly large number of cases, the honest answer is “nothing at all.”

We can call it “cheating.” Or we can call it reality.

AI didn’t break the centuries-old system but rather revealed that the system was flawed due to one fragile assumption: producing text equals producing thought.

No, it doesn’t. Not anymore.

AI Is Rocket Fuel for the Curious  –  and a Crutch for Everyone Else

Here’s where things get really uncomfortable.

AI doesn’t make everyone smarter. It amplifies whatever is already there.

Give a bright, curious student access to ChatGPT and you’ve effectively granted them superpowers. They can dive deep into any topic, get instant feedback, test ideas in real time, and iterate so fast that it would have looked unhuman ten years ago.

Give a student who doesn’t care the same tool and they will use it to avoid thinking altogether. Why sweat over an assignment when you can outsource it to the machine and watch Netflix instead?

So we don’t just get a bigger gap in results. We get a split in the fundamental process of learning:

• One group uses AI to think with it – they treat it as an extension of their mind.
• Another group lets AI think instead of them – they use AI as a substitute for their mind.

Trying to educate both groups with the same model  –  same classroom, same assignments, same expectations  –  will feel increasingly absurd. You just don’t teach 17-year-olds how to drive on a race circuit in Monaco along with the real F1 pilots.

The Teacher Is Not the Smartest Person in the Room Anymore

If AI can out-explain you on your own subject in 10 seconds, your job description as a teacher changes.

As sad as it sounds, we no longer need the “sage on the stage”  –  the walking textbook at the front of the room. That job has gone to the machines.

In the world of AI, the teacher who survives the transition does something else.

The job of a modern teacher is to make students care, make students passionate about the subject  –  if they can. They must teach students how to think, which is now more important than what to know  –  how to ask the right questions, how to spot nonsense and work around the imperfections of AI, how to tolerate confusion and information overload. They become guides to AI itself, coaching students on how to use AI as a microscope rather than a vending machine. And they act as curators in a world drowning in information (and misinformation), helping students detect what’s signal and what’s noise.

Almost no teachers are trained for this, and many are running in the wrong direction. They don’t use AI, they’re suspicious of it, some ban it. I’ve seen the Instagram reels: red-faced professors in auditoriums, literally losing it at their students, yelling “ChatGPT is not allowed in my class” and “I will not tolerate essays written by a polite robot.” It’s the theatre of denial. In 2025, banning AI in a classroom is not pursuing academic integrity; it’s willful professional negligence. They are training kids for a labour market and a cognitive environment that doesn’t exist anymore.

AI is here to stay  –  which is more than you can say about the job security of these AI-intolerant professors.

The G.O.A.T. Effect: Why Masters Become More Valuable, Not Less

If AI can teach almost anything to almost anyone, you would think the value of human teachers goes to zero.

Nope. Not at all.

In reality, AI destroys the average, but it makes the exceptional a lot more valuable.

There will always be enormous demand to learn directly from people who actually made it to the top. Yes, that might mean learning cinematography from an Academy Award winner, but it also means learning product vision from the immigrant founder who arrived with nothing, built a real company, and took it public. Gen Z even has a term for that  –  G.O.A.T., the Greatest Of All Time.

AI does know the material, but it can’t fake the scar tissue. It can’t teach you the things that aren’t in the textbooks  –  the bad bets, the near-firings, the 2 a.m. decisions that quietly separated a “promising career” from “I made it to the top.”

So as AI pushes the price of average teaching toward zero, the price and demand for GOAT-level human teaching will go up.

The New Stack: AI, Mentors, Masters

At the base, AI will soon become the default teacher for generic knowledge. The jury is in on the fact that an AI tutor will be able to bring any student in the world from zero to baseline competence in an astonishing range of fields.

On top of that will sit the human layer: the people who take over what we used to call “teacher.” Their job is not to out-lecture the AI; it’s to coach you on how to get the best out of it and to keep you paying attention. They add context, accountability, nuance. They teach you how to think with AI and stop you from letting it think instead of you  –  how to ask better questions, how to smell nonsense, how to push back on the machine when it sounds confident but wrong, a progressively critical skill set these days.

Above that, for a smaller, elite group willing and able to pay, is the master tier: the GOATs. This is where learning looks more like an apprenticeship or a dojo than a classroom.

Compare that to a traditional university: one human at the front trying to be all three layers for 200 students in a lecture hall.

It’s not even a fair fight.

So What Happens to Schools and Universities?

This stopped being an abstract question for me this year, when my daughter started college. Suddenly, it wasn’t “the higher-ed debate,” it was an invoice for roughly $50,000 a year. And I found myself asking the only honest question a parent in 2025 can ask: what exactly am I paying for?

For the good part of the past 100 years, the answer was a bundle: you paid for knowledge (the lectures and curriculum), community (the people and the experience), credential (the piece of paper), and brand (the name on that paper and on your LinkedIn).

AI blows a hole straight through the first one. In providing knowledge, the university no longer has any real advantage. That doesn’t make universities irrelevant, but it does mean knowledge alone is no longer what justifies $50k a year.

Where universities can still create real value  –  and where they’ll live or die over the next decade  –  is in the other three: community, credential, and brand.

Community is the four-year human network marathon: the friends you make, the roommates who become co-founders, the late-night arguments, the social capital you build. Credential is the filtering function: the signal to employers and investors that you got through a certain set of challenges and survived. Brand is… well, people don’t buy Gucci T-shirts because at $500 they’re objectively 20x better than Uniqlo. So, Harvard isn’t going anywhere.

Those three are not trivial. They are, in many cases, worth paying for. But once AI has made the knowledge essentially free, universities have to work much harder to earn the rest of the price tag. Some will double down and actually build extraordinary communities and stronger brands. Many others will keep their head in the sand, pretending “knowledge” is still the reason students should keep paying them.

Rebuild, Don’t Retrofit

The most dangerous reflex in education right now is the instinct to “add some AI” to the existing centuries-old structure. A chatbot here, a plagiarism detector there, a unit on “AI literacy.”

That’s not what we need. It is painful watching us trying to breed a faster horse for the world that wants a car.

We should stop trying to sprinkle AI onto a dead body, hoping it will walk again. AI can seem like magic, but this aint Hogwarts. We must start from the assumption that every student has a superhuman, tireless tutor in their pocket at all times  –  and design backwards from there. Assessment has to shift from take-home artifacts to visible thinking, live debates, collaborative work, and real-life projects where you can’t fake the journey. Teacher training has to shift from “know your subject” to “know how humans learn with machines.”

And for high-agency students who actually want to learn we must build education tracks that look less like school and more like a startup accelerator.

The Real Divide of the 21st Century

The biggest inequality in the age of AI will not be who has access to the tools. The tools will be cheap and abundant.

The real intellectual moat will be between people who learned to think with AI and people who are allowed to let AI think for them.

The first group will be superhuman by historic standards  –  AI will multiply their intellectual capabilities and productivity to a cosmic proportion. The second will end up being highly dependent, easily manipulated, and very confused by a world that moves too quickly and speaks a language they never really learned.

It is the modern educators’ job to decide which side of that line most of our kids end up on.

Right now, by banning AI and defending a model built for another century, we’re making that decision by default.

The countries and the societies that have the guts to rebuild from the ground up will own the future. The rest will become exhibits in the museum.