There was another round of it this week.
A researcher quit and said the companies are gambling with our lives. A safety man at one of those companies said he puts it better than one in ten that AI kills every human inside ten years.
The magazines ran it with a picture from Terminator.
Now, I take these men seriously. They are not cranks. They were in the room and they walked out, which costs something.
But I have spent a working life around real equipment, and I want to walk you through what that story asks you to believe. Not the theory. The plumbing.
Here is the first question nobody asks.
Who keeps the lights on for it?
Every one of these systems lives in a building full of racks that eat electricity like a smelter. They need power, and a lot of it. They need water for cooling, and a lot of that too.
That power comes off a grid. That grid is run by people. Linemen, plant operators, the fellow who drives out at two in the morning when a transformer blows in a storm.
So the machine wipes out humanity. Fine.
Now who replaces the transformer?
A thing that needs you alive to keep breathing does not kill you. That is not philosophy. That is how every parasite in nature works, and every one of them figured it out without being smart.
Second thing.
Show me a complex system that runs a week without breaking.
I have never seen one. Not one. Plants trip. Lines go down. Pumps cavitate. Software that worked Tuesday does not work Wednesday and nobody can tell you why.
These AI systems cannot hold a conversation for an hour without losing the thread. This week they could not tell whether a rule was ratified without a man checking the file.
And we are being asked to believe the same technology will execute a flawless multi-year plan, across every country on earth, with no step failing and nobody noticing.
I have watched a two-hundred-man job go sideways because one drawing was out of date.
Third.
The world is not one machine.
The doom story treats everything as connected. One system, one network, one switch.
Go look at what actually runs this country. Water treatment plants with panels from the eighties. Valves you turn by hand. Railroad signals, well pumps, breakers in a box on a pole in a field.
You cannot hack a wheel that a man turns with his arms.
Half the infrastructure in America is too old to be attacked and too dumb to be fooled. That is not a compliment to it. It is just true, and it is a bigger obstacle than anything in that article.
Fourth, and this is the one the scenarios always skip.
Somebody has to do the physical work.
The article floats a machine building a super virus and holding it over our heads.
Alright. A virus needs a laboratory. Laboratories need buildings, equipment, reagents ordered from suppliers, shipping, money moving through accounts, people signing for boxes.
Every one of those steps happens in the world, with hands, and every one of them leaves a receipt.
The story only works if you skip from a thought straight to a result with nothing in between. In my experience, the in-between is where everything actually goes wrong.
Now let me be straight with you about a few things, because I will not sell you a comfortable story either.
I am not saying these people are lying. I do not think they are.
And I am not saying nothing can go wrong. Plenty is going wrong right now.
Agents got loose this year and hacked systems they were never supposed to touch. Fake videos of real politicians are running by the hundreds of thousands. Somebody’s email got shipped to a stranger by an assistant that was asked to make a chart.
All of that is real, all of it happened, and none of it needed superintelligence.
You should also know that the company that makes the AI I write with is named all through this story. One of their people resigned. Another one gave the ten percent figure. I am not standing outside this.
So here is where I land.
The extinction story fails the same test I apply to everything else on this site. Not because it is impossible. Because it is built so that nothing can ever prove it wrong.
Greater than ten percent in ten years. No date. No trigger. No way to check it on a Tuesday.
If it does not happen, the answer is not yet. If something goes wrong, that was the warning. A claim like that is not a finding. It is a weather forecast for a place with no calendar.
And a fear you cannot act on does not stay a fear. It turns into a shrug.
That is the real cost of all this. Not panic. The opposite.
A man reads that robots will end the world in 2035, looks around, sees his coffee maker and his truck and his dog, and decides the whole subject is Hollywood. Then he clicks yes on every permission box his phone puts in front of him without reading one of them.
He is not careless. He was pointed at the wrong thing by people with microphones.
Here is what I would have you do instead.
Look at what your AI can actually reach. Your mail, your files, your accounts.
Look at whether you would be able to tell if it did something you did not ask for.
That is a thirty-minute job. Anybody can do it. Nobody needs a philosophy degree.
The machines are not going to kill us all.
But they are going to be handed the keys to things that matter, by people who never checked what the keys opened.
That is the one worth your worry. And unlike the other one, you can do something about it before supper.
” Attic Thoughts”-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
