Learning is an asset that yields without being used up

Translated from the original Spanish article.

Something curious happens when a machine becomes an oracle. Someone who knew nothing about a subject on Monday consults a chatbot on Tuesday, and by Wednesday is explaining it to you—you, who have spent years on it and happen to know a thing or two about it.

They haven't learned anything. They've been told. The errand boy of knowledge. Like the pilgrims who climbed up to Delphi, they come back with the answer in hand, ready to brandish it, without the slightest curiosity about how it was reached.

The current debate about artificial intelligence revolves around productivity. But always from an industrial point of view. The question is whether the machine makes accounting faster, lawyers more efficient, the process lighter. It's a legitimate question; I take it on too. It's as fun as it is stimulating.

But, since we're being productive, AI has an obvious virtue: it can help us learn. And how, let me tell you!

Of course, that's more boring. And it takes effort, too.

So that I can be understood in today's conversation, I'll put it in its own financial-business terms: Learning is acquiring an asset that yields without being used up (as long as our heads hold out).

You use what you know again and again, and you lend it at no cost to yourself. That's leverage. And since each thing learned makes the next one easier, or richer, the returns compound. Compound interest is the silent force of a mind.

Seen this way, whoever turns to the oracle of Delphi, Gemini, Claude, or whichever it may be, has it exactly backwards. They use the most extraordinary learning device ever built in order not to learn, to close a question instead of opening it. They've borrowed a conclusion they can't sustain, which is a tolerable description of recklessness, or of foolishness, as the case may be.

Beside them, the colleague who uses the same tool to check, to replicate, to follow a thread they couldn't have followed alone, compounds little by little. In a year, the gap between the two will be small. In ten, it will be an unbridgeable difference.

The productivity of artificial intelligence in a company's processes will come, in its own time. Its capacity to compound the knowledge of a company's professionals, however, is already available, to whoever knows how to resist treating it like Delphi. Make the most of it, and gladly, because that disposition, judging by the evidence, is in short supply.