The question with your name on it

You have spent hours this month trying to find out where AI is going, and you still do not know whether to open the course tonight.

That is not a failure of research. It is a question doing the work of a different question, and it is worth separating them slowly, because the confusion is doing real damage to people who are being sensible.

Two questions, one of them yours

Whether to keep going is a question about a decision. It has somebody who decides, and that is you. It has a deadline, and that is tonight, because tonight you either open the thing or you do not.

Whether progress is slowing is a question about the world. Nobody decides it. It has no deadline. It will be settled over several years by events, and by people who have never heard of you.

Only one of those is yours, and it is not the one that gets typed into a search bar.

Nobody swapped them on you deliberately. The second sounds more like research and less like nerves, so it is the one that gets asked, and because it is a genuine factual question, everybody answers it. Sincerely, at length, from both directions. They are being helpful. They are answering what you said.

Which is why more reading resolves nothing. The fault sits upstream of every article you have opened: somebody has to swap the question back, and only one person is in a position to do that, because only one person knows which question they were reaching for.

The replacement

Ask this instead, of anything you are about to spend months on:

Would this still be worth having if I turn out to be wrong about AI?

Sit with it a moment before answering, because the first answer is usually about the tool and the useful answer is usually not.

Some things clear that bar comfortably. Reading a system somebody else built and working out what it is doing. Saying out loud what you chose and what you gave up. Naming what a trade-off actually costs. Building things that survive a change underneath them.

None of those is a bet on a direction. Each pays in the fast version of the next five years, and pays in the slow one. That is the only property worth selecting for while nobody can tell you which version you are in.

What it costs, said plainly

This is slower than being right.

If you pick the direction correctly and commit everything to it, you arrive before the person who spread their bet. That is arithmetic, not rhetoric, and pretending otherwise would make the rest of this less trustworthy.

So the real comparison is a faster route that requires you to be right about something nobody knows, against a slower one that does not. Given that nobody does know, the second looks like the better trade. But it is a trade, and you should make it with your eyes open.

There is a quieter cost too. The things that survive either outcome tend to have no certificate attached, so progress on them is harder to show and harder to feel. Expect months with nothing to point at.

One way to check the whole idea

Take something you learned years ago and still use today. Ask why it lasted.

Almost never because somebody predicted the direction correctly. Almost always because it would have been worth having whichever way it went.

That is the test, applied backwards, and it is the reason to trust it forwards.

The four-column page for running this on the thing you are actually deciding is here, and the middle two columns are the ones that decide it.

Where certainty is not available, the move is to stop pursuing it and start picking things that never needed it.