AI Can’t Hang Your Door

I got hung up today. I had a door to put in, and it ate most of the day, the way a door will.

The whole country’s watching the news, and I was on my knees with a handful of shims. I’m not sorry about it, because that door taught me something I can use with you.

A door doesn’t care what anybody thinks of it. You can have the best plan in the county and the nicest door money can buy. If the jamb is out of plumb, that door will tell you. It’ll swing open by itself, or drift shut, or rub at the top corner. It has no opinion and no politics. It just reports what’s true.

That’s why you don’t hang a door by feel. You put a level on the hinge side first, and you shim behind the hinges until the jamb stands straight. You check it, then you close the door and look at the gap all the way around. If the gap’s even, you’re getting somewhere. Then you open the door halfway and let go. If it stays where you left it, you’ve done your job.

Now let me tell you the part most people never think about. A level can be wrong too. Drop one on a hard floor and it may never read true again. So you test it. Set it against a surface, read the bubble, then flip it end for end in the same spot. If the bubble reads the same both ways, you can trust it. If it doesn’t, the tool has been lying to you, and it would’ve kept on lying with a straight face.

You see where I’m going with this. A lot of you have an AI on your phone now. You ask it questions and it answers fast, in nice clean sentences. It sounds as sure of itself when it’s wrong as when it’s right. There’s no wobble in the voice and nothing to warn you. That answer is a bubble in a level you haven’t tested yet.

I don’t say that to scare you off. I use AI every day, and it’s helped me on jobs around this house. This very morning it gave me solid advice on a building question. Then I pushed back with something I knew from the trade, and it had to admit my point was a good one. It still had a good reason for the way it wanted the job done, and I went with that. But that only happened because I didn’t just take the first answer and start drilling.

That’s the whole trick, friends. You’ve got to flip the level. When an AI tells you something that matters, ask it where that came from. Ask it what it isn’t sure of. Ask it the same thing a different way and see if the answer holds. Then go check it against something that isn’t the AI, like the label, the manual, the doctor, or the fellow at the counter who has done this for forty years.

Most of all, check it against the job itself. The machine can tell you how a door ought to hang. It has never stood in your hallway. It doesn’t know your floor runs low on one side, or that some fellow framed that wall in a hurry sixty years ago. You’re the one standing there with your hands on it. You can see what it can’t. That makes you the one in charge, and I wouldn’t hand that job off to anybody.

The more a thing costs if it’s wrong, the harder you check it. If an AI gives you a bad recipe, you’ve lost a supper. If it steers you wrong on your medicine, your money, or what’s holding up the roof, you could lose a good deal more. You wouldn’t let a stranger’s word decide those things without looking. A machine’s word deserves the same look, no matter how smooth it sounds.

There’s something else that door reminded me of. Nobody can do your checking for you. No company is going to do it, and no government either, at least not at your kitchen table when you need the answer. They may write rules, and I hope they write good ones. But the last check happens where you’re standing, with you and whatever tool you’ve got.

It’s not hard to do, and you don’t need a degree. All you need is the habit a carpenter has, which is to measure, look, and test before you trust it. You’ve probably done that your whole life with everything but this. All I’m asking is that you bring it along when you sit down with the machine.

The door’s in now. It opens, it closes, and it stays where I leave it. The level says so, and I’ve checked the level.

Go check yours.

Load this file before each AI chat session.

the-faust-baseline-public-release-1Download

This post was drafted with AI governed assistance and reviewed and directed by Michael S. Faust Sr. before publication.

” Attic Thoughts”-library – Intelligent People Assume Nothing