The Best Question Beats the Right Answer

Every test your kid has ever taken has the same shape. Someone else holds the questions. Your kid supplies the answers. We've run childhood this way for a hundred and fifty years, and it made a kind of sense — answers were scarce, and the people who carried them around in their heads were valuable.

Then, in about two years, answers became the cheapest thing on earth.

Think about what actually happened. A machine now exists that produces competent answers to almost any question, instantly, for roughly the cost of the electricity to light a bulb. Whatever your feelings about that machine — and this talk is not asking you to love it — one economic fact is already settled: the price of a decent answer has collapsed. And when the price of something collapses, value doesn't disappear. It moves. It moves to whatever is still scarce.

What's still scarce is the question.

Here's the thing nobody tells kids: the machine is a mirror with a delay. Ask it something dull and it hands you back polished dullness. Ask it something sharp and it becomes, quite suddenly, the most useful colleague you've ever had. The quality of what comes out is chained to the quality of what goes in — which means the person asking has become the ceiling on the whole system. Not the model. The asker.

We have a name for this in the industry — prompt engineering — and I half wish we didn't, because the name makes it sound like a technical specialty, like plumbing. It isn't. Strip away the jargon and what remains is the oldest intellectual skill on record. Twenty-four centuries ago, a man wandered around Athens doing nothing but asking questions, and the questions were so well-built that we're still assigning the transcripts. Socrates never lectured. He asked. And his questions did something answers can't do: they made the other person think things they didn't know they could think.

That's what a great prompt is. It's not magic words. There are no magic words — I want to say that plainly, because kids trade rumored magic phrases the way they trade cheat codes, and it's a dead end. A great question is built, and it's built from parts you can name.

It has context: here's who I am and what I already know. It has stakes: here's what this is for, here's what happens if it's wrong. It has constraints: explain it in the length of a bus ride, use no jargon, assume I've never seen a spreadsheet. And it has a follow-up waiting — because real questioners never ask just once. Watch a kid go from "tell me about volcanoes" to "I'm building a warning system for a town near a volcano — what's the earliest signal it would need to detect, and how would I fake that signal to test it?" That second question isn't a school exercise anymore. It's a project. The machine's answer to it is the beginning of something, not the end.

And here is the part I find hopeful, as someone who spends a lot of time around kids and machines: question-asking is learnable the way throwing is learnable. Nobody is born with it. It grows with repetitions. But — and this is the design flaw worth saying out loud — school gives kids almost no repetitions. A researcher once tallied who asks the questions in a typical classroom, and it isn't close. The teacher asks hundreds per week. Each student asks a handful, and most of those are logistical. Can I go to the bathroom. Is this on the test. Somewhere between kindergarten and middle school, the question-asking muscle goes quiet, not because kids run out of curiosity but because the room stops rewarding it. There's no column in the gradebook for asked something nobody thought to ask.

So the muscle has to get built somewhere else. It gets built wherever a kid is working on something they actually want to exist — a game, a chatbot with a personality, a comic, a gadget — because want is where real questions come from. A kid who wants their robot to stop falling over does not need to be taught to ask better questions. They need a room where asking is the normal thing happening, other kids doing it alongside them, and a machine patient enough to answer the first clumsy version and the fifth sharper one. The machine's patience, honestly, is its most underrated feature. It has answered ten thousand bad questions before breakfast and it will never once make your kid feel stupid for asking an eleventh. No human teacher can offer that. It changes who's willing to raise a hand.

I should be honest about the objection, because it's a good one. Isn't this just kids getting answers handed to them? Isn't the machine doing the thinking? Here's what I've seen, and I've now watched it hundreds of times: handed answers are inert. A kid who receives an answer they didn't really ask for scrolls past it, the way you scroll past a billboard. But an answer to a question they built — built with stakes, for a thing they're making — gets used. It gets argued with. It gets tested against reality, because the robot either stops falling over or it doesn't. The question is what turns an answer from content into material.

The kids who figure this out are going to be very hard to compete with, and I don't just mean in careers, although I mean that too. Every field is reorganizing around the same division of labor: machines supply answers, humans supply the questions worth answering. The scientist's edge is no longer knowing the literature — the machine has read all of it. The edge is asking the question the literature hasn't thought of. Same for the founder, the doctor, the reporter, the twelve-year-old with a strange idea for a game.

For a century and a half we sorted children by how well they answered. The sorting has flipped, quietly, in under a decade — and almost nothing about how we raise kids has flipped with it yet.

Which means the families who notice now, notice early.

So here's my homework, and it's for the adults. Tonight, when your kid tells you about their day, skip what did you learn — that's an answer question, and it gets answer-shaped shrugs. Ask instead: what did you ask today? First night, you'll get a blank look. Ask again tomorrow. The night your kid comes to dinner with a question already loaded, because they know someone's finally going to ask for it — that's the night the muscle starts growing back.

Answers just got cheap. Raise an asker.