You have probably felt this already. You give an AI tool a rough idea, and it hands back work that used to take a small team a week. You also give it a half-baked, rushed thought, and it hands back a polished version of that same rushed thought. Bigger, more sure of itself, and harder to undo than what you started with.
A few years ago, intelligence was expensive. Now it is almost free, and it gets cheaper every few months. That changes the job. Here is what I think the new job is.
The problem everyone is talking about
Peter Diamandis wrote about the next five years in a piece called The Next 5 Years: A Supersonic Tsunami. His main point is that AI is getting cheap fast. A few years ago, asking an AI a hard question could cost about twenty dollars. Now the same question costs a few cents, per the Stanford AI Index 2025. And it keeps dropping, about ten times cheaper every year.
When intelligence gets that cheap, the slow part moves somewhere else. Diamandis puts it simply: intelligence used to be rare, now it is everywhere. The slow part is now the real world. Tests still take time. Factories and machines still cost real money. Demis Hassabis has even said we might cure most diseases within about ten years, once cheap intelligence is aimed at biology.
Take those timelines with a grain of salt. The people saying them want them to be true, and they have been too hopeful before. But the direction is real, even if the dates are off.
The bigger problem underneath
There is a second shift in the same piece, and Diamandis only touches on it. When he says what to learn now, he says: get good taste, get good judgment, and learn to ask the right question. None of those are tech skills. They are about choosing well when intelligence is no longer the hard part.
That is the part I keep thinking about. Free intelligence does not change the person using it. It makes that person bigger.
If you are stressed, scared, and trying to look impressive, AI makes all of that bigger. A hundred fast moves an hour, each one louder, and none of them really yours. If you are calm and clear about the work that is actually yours to do, AI makes that bigger instead.
Intelligence on its own is not the edge anymore. You also have to be a person who can aim it well. Whoever you already are, AI turns the volume way up.
What I mean by alignment
I do not mean AI safety. I mean something simpler and personal.
The common advice stops at taste, judgment, and good questions. Those are real, and they last. But they sit on top of something people skip. Good taste comes from being calm enough to sit with a hard problem. Good judgment comes from practice, enough that it holds up when things move fast. Good questions come from actually knowing what you are trying to build.
So it comes down to three plain things: staying calm, having judgment you have practiced, and knowing what you actually want. That is the base the lasting skills grow from. You cannot buy it with more intelligence. It is slow work, and the AI cannot do it for you.
Who this will not help
If your plan is to stay stressed and just bolt on AI to go faster, this makes things worse, not better. More speed on a shaky base means more wrong turns, made with more confidence, and harder to fix once they spread.
The people who get the most from AI will not be the ones with the most of it. Everyone will have that. It will be the ones who know what to aim it at, and who stay steady enough to keep aiming.
Where I have landed
This is close to something I wrote a couple of weeks ago: the best thing any inner-work practice can do is help you get clear on what you are here to do, and then do it. That post is building inner work that actually delivers.
So here is my bet, and it is a bet, not a sales pitch. The change is real. I think the hard part moves to staying calm and clear, and that is what I am building on. I could be wrong about how fast it comes, or how many people will need it. Ask me in a year.
