The Floor and the Legs

Afriend’s son, 16, told me over dinner that he no longer reads the books his school assigns. He reads the machine’s summary of them, then asks the machine what a clever student would say, then writes a version of that down in his own hand. He gets A’s. He finds the arrangement faintly embarrassing, the way you’d be embarrassed to be caught taking the lift up a single floor. But when I asked him to defend the stairs, he couldn’t. For a moment, neither could I.

This is the question now sitting underneath all the others. If the finished product of an education (the essay, the analysis, the diagnosis, the working line of code) can be summoned in seconds by anyone with a phone, then what, exactly, is the education for? Why spend four years and a mortgage’s worth of money acquiring something the machine will hand you at the drop of a hat, fully formed and lightly polished, indistinguishable at a glance from the real thing?

It is a serious question, and the people who wave it away are not paying attention. The tool is genuinely good. It does not merely look up facts the way an encyclopaedia did. It writes the paragraph. It builds the argument. It performs the one act that, for most of recorded history, we treated as the proof that learning had occurred: you demonstrated that you understood a thing by making something new out of it. Now the making is free.

And it is frictionless, which is the part that does the damage. Learning, as anyone who has done it knows, is mostly friction: the dull resistance of a problem that won’t yield, the hour you don’t understand the chapter, the draft you throw away. The machine removes the friction entirely. The answer arrives whole, calm, expert, and the small struggle that used to leave a residue in the mind simply never happens. You get the floor without the climb. You feel, briefly, like you have arrived somewhere.

Every generation has had its version of this panic, and they make for reassuring company. In the Phaedrus, Socrates tells the story of the god who invents writing and the king who refuses the gift, warning that it will hollow out memory and leave men with the appearance of wisdom in place of the thing itself. The calculator was going to end arithmetic. Wikipedia was going to end scholarship. Each time the discipline survived, the frontier moved, and the doomsayers were filed under Luddite.

I would like to offer that comfort. I don’t think it is quite available this time, and the reason is precise. The earlier tools all replaced storage: memory, lookup, the holding of facts. None of them touched judgement. This one does. Or rather, it does something stranger: it produces a flawless imitation of judgement, and produces it most confidently at the exact moment it is wrong. The old king was half right after all. The danger was never the loss of memory. It was the appearance of wisdom worn by people who have none.

Here is the thing the 16-year-old can’t yet see. The machine does whatever you ask of it precisely as well as you know how to ask. To a trained mind it is an instrument of rare power. To an untrained one it is a fluent, tireless, supremely confident liar. The untrained mind, by definition, cannot tell which of the two it is holding. The tool does not close the distance between the expert and the novice. It widens it, and then hides the widening behind output that looks identical coming from either.

I spend my working life around radiologists, so let me borrow their world for a moment. An algorithm can now flag a nodule on a chest scan faster than a tired human at the end of a long list, and in narrow tasks more reliably. Good. But the flag is not the medicine. Whether this nodule, in this patient, with this history, read against a prior from 2022, means anything at all: that is judgement, and judgement is what 13 years of training and 10,000 scans actually bought. The tool reads the image. The radiologist reads the situation. Remove the second and you have not saved money. You have built a very fast way to be wrong, and dressed it in the language of progress.

So the future of education is not redundancy. It is something quieter and more dangerous: invisibility. We are about to lose the ability to tell, from the outside, the difference between a person who knows and a person who can merely fetch. The two will submit the same essay, ship the same code, give the same assured answer in the same meeting. The difference will surface only under load: when the tool is unavailable, or subtly wrong, or being used against you by someone who understands it better than you do. By then it is too late to start climbing.

Which returns me to the boy and his stairs. He had the metaphor backwards, and so do most of the people arguing about any of this. He thought the point of the stairs was to reach the floor, and once a lift exists, only a sentimentalist takes the stairs. But the climbing was never about the floor. It was about the legs. The educated mind is not a warehouse of retrievable facts that the machine has now made surplus. It is the one instrument capable of knowing whether an answer is any good, and that instrument is built only by friction, only by the climb, only by doing the work the machine now offers to do for you.

We were never teaching people the answers. We were building the thing that can tell a good answer from a confident one. It has just become the most valuable possession a person can own. The trouble is that, for the first time, it has also become optional. And almost no one is choosing it.