Everyone Has a Private Tutor Now. Most People Use It Wrong.
AI can be a private tutor or an answer vending machine. How to use it to do the thinking yourself, and why learning together in person still matters.
Audio version
Listen to this article
Narrated by Microsoft Foundry.
For most of history, the best way to learn something was to have someone sit next to you. Someone who knew the subject, knew you, answered your questions at your level, and let you try again when you got it wrong. Kings had that. Rich families had that. The rest of us had books, classrooms, and if we were lucky, one patient senior colleague.
That just changed. Everyone with a phone now has a tutor available at two in the morning, in any subject, with infinite patience and no judgment. It is the biggest shift in how people learn since the printing press, and most people are using it as a vending machine.
The vending machine mistake
Here's how most people use AI to "learn": they have a problem, they describe it, they get an answer, they copy it, and they move on. Problem solved. Nothing learned.
That's not learning. That's outsourcing. And it's a trap, because it feels productive. You finished the task faster than ever and you understand it less than ever. Do that for a year and you end up in the place I warned my sons about in my keynote: weak, with excellent tools.
The difference between a vending machine and a tutor is not the machine. It's how you ask.
Ask like someone who wants to learn
A good tutor doesn't hand you the answer. A good tutor asks what you already know, explains at your level, lets you attempt it, corrects your attempt, and pushes you one step further than you'd go alone. You can get every bit of that from AI, but you have to ask for it. A few patterns that changed how I learn:
"Don't give me the answer. Ask me questions until I get there." The single most useful sentence. It turns the model from an answer machine into a Socratic tutor, one that teaches by asking questions and makes you do the thinking.
"Explain this as if I'm a .NET developer who has never touched Python." Tell it what you already know. A tutor who knows your starting point skips what you don't need and slows down where you do. No course can do that for a room of thirty people. A tutor can do it for you.
"Here's my attempt. What's wrong with it, and why?" Try first, then ask. The learning is in the gap between your attempt and the correction. Skip the attempt and you skip the learning.
"Quiz me on this tomorrow." Or today. Make it test you. Being asked a question you can't answer is uncomfortable, and that discomfort is exactly where memory forms.
Notice what all four have in common: in every one, you do the work. The AI shows the path. You still walk it.
Why this is better than what we had
I don't say this lightly, because I run a community built on rooms full of people learning together. But as a tutor, AI beats most of what came before, for three reasons.
It has no face to lose. The question you were too embarrassed to ask in a classroom, the one you've been told is too basic? You can ask it ten times, ten different ways, and the tutor never sighs. For a lot of people, that alone unlocks subjects they'd given up on.
It goes at your pace. Not the pace of the fastest or slowest person in the room. Yours. Stuck on step three? Stay on step three for an hour. Already know steps four through nine? Skip them.
And it's there when the learning actually happens. Nobody learns on a schedule. You learn when the problem is in front of you and you're motivated to beat it. A tutor at that exact moment is worth more than a course next Tuesday.
In-person learning vs AI learning
So does the tutor in your pocket replace the room full of people? I've thought about this more than most, and the honest answer is that they're good at different things.
What AI learning wins at. The three things above, which all come down to one word: explanation. For understanding a concept, step by step, at your own speed, nothing we've ever had comes close.
What in-person learning wins at. Everything that isn't explanation.
Motivation: you show up because other people are there, and you finish because they're watching. Nobody has ever been held accountable by a chat window.
Calibration: you see that the person next to you is also stuck, and suddenly you're not stupid, you're learning. The tutor can tell you that. The room shows you.
Teaching: the moment you explain something to the person beside you, you understand it twice as well. There's nobody to explain things to in a chat.
The person three steps ahead: not the expert on stage, but the one at your table who did this last month and remembers exactly where it hurt. That person is the best teacher in the building, and no model knows they exist.
And trust over time: a mentor who's known you for years notices what you avoid, what you're ready for, and when to push. The tutor in your pocket starts every conversation fresh.
And where AI loses outright. It can be confidently wrong. A human who doesn't know says "I'm not sure." The AI sometimes says a wrong thing beautifully. So verify the way developers always have: build it, run it, break it. Your compiler doesn't hallucinate.
Here's how I'd put the split in one line:
AI is better at helping you understand. People are better at helping you become.
The concept you can learn alone at two in the morning. The developer you become, you become in rooms.
I've watched a year of free conferences fill up while every answer was online. People didn't come for answers. They came to learn together. So the new way of learning isn't tutor instead of room. It's tutor plus room. That's exactly why our hands-on workshop format puts people together to build something, with AI next to each of them. The tutor does the explaining. The room does the rest.
The only rule that matters
If I had to give one rule to anyone, from my own sons to a senior engineer learning a new stack, it's this: AI may explain. It may not do your learning for you.
Because learning is the one thing you can't automate. You can only skip it. The task can be done by the machine. The understanding has to be built by you, one attempt and one correction at a time. That was true with a tutor next to you in a palace three hundred years ago, and it's true with one in your pocket today.
Everyone has the tutor now. The question is who's willing to be the student.
Pass it on