The headline said one word settled the whole argument about whether artificial intelligence is a bubble.
One word. Said one time. On an earnings call.
The word was visibility.
The idea was that the head of the company can now see further down the road than he ever could. He can see the land. The power. The parts. The buildings that don’t exist yet.
And because he can see it, the writer said, the worry is over.
I don’t know if there’s a bubble. I have no idea what that stock is worth. I’m not going to pretend otherwise, and if that’s what you came for, this isn’t it.
What caught me was something else.
A man said he could see. And a writer took that as proof he could see.
That’s the whole move. Somebody with money riding on the answer said the answer, and the saying of it was counted as the evidence for it.
I’ve spent a year and a half on this exact problem, in a different room.
I work on how you hold an artificial intelligence to a standard. And the first thing I had to write down, before any of the rest of it, was this: declaring something is not doing it.
Any machine can tell you it checked its sources. Saying it checked is not evidence it checked. The words cost nothing. A system that lies and a system that doesn’t say the same sentence.
So the rule I built everything else on is simple. A claim only counts if holding to it leaves a mark somebody on the outside can look at.
Not a promise. A mark.
Now here’s why that story got worried me.
The man on the earnings call didn’t leave a mark. He said a word. The company put out a forecast that reaches further into the future than it usually does. And that forecast was treated as the proof that the demand behind it is real.
But the forecast came from the company selling the thing.
That’s not a check. That’s the same voice, saying the same thing, twice.
I’m not calling anybody a liar. He may be exactly right. He may see all of it, clear as morning. I have no way to know, and that’s the point I keep coming back to.
There’s nothing in that story I could look at and say — here, this is how we’d find out if he’s wrong.
That’s the test. Not “is he confident.” Not “does he sound like he knows.” The test is whether there’s anything in front of me that could come back and contradict him.
If there isn’t, I don’t have evidence. I have a man’s word and a nice feeling about it.
I’ve been building rules for machines this whole time. Rules about naming your source. Rules about stopping when the evidence stops. Rules about not sounding more sure than you have grounds to be.
I thought I was writing about artificial intelligence.
Turns out I was writing about something older than that.
The market does it. The news does it. Every one of us does it, sitting at a kitchen table, saying a thing with enough steadiness in our voice that nobody asks what it’s resting on.
Confidence is the cheapest thing there is to produce. It always has been. The machine just made it cheaper, and faster, and put it in front of more people at once.
So here’s what I’d leave you with.
The next time somebody tells you the argument is settled, don’t ask whether they sound sure. Sure is free.
Ask what would have to happen for them to be wrong. Ask what you could go look at.
If the answer is that the person who benefits said so, you don’t have an answer. You have an advertisement with better manners.
I still don’t know if there’s a bubble.
Neither does that article. That’s what I actually learned reading it.
CHALLENGE: Pick one thing you’re certain about today. Just one. Now name the thing that would prove you wrong. If nothing comes, you don’t hold a conclusion — you hold a habit.
Post Library – Intelligent People Assume Nothing
Contact: micvicfaust@gmail.com
This post was drafted with AI governed assistance and reviewed and directed by Michael S. Faust Sr. before publication.
© 2026 The Faust Baseline LLC | All Rights Reserved
