What the Machine Was Never Told to Look For
Subscribed
_
This is another insight which I had partially finished for several weeks. In The Fish in the Canoe I glanced off AI, and this piece presented itself as the next step in that particular chain.
_
In the early 1980s I worked in the semiconductor industry, in a building where the air was cleaner than an operating theatre’s and nobody ever raised their voice.
The machines there had a form of artificial sight. They joined a wire a third the thickness of a human hair to a pad on a silicon chip, hundreds of times a minute, and they did it by pattern: a stored picture of how the thing should look, checked against how it did look. It was, by the standards of the day, miraculous.
And the most reliable check on the whole of it was a person at a microscope, looking.
The machine was excellent, and it never tired. It was excellent at finding faults it had been shown.
What it could not do was be puzzled.
The person at the microscope would stop at something that was on no list: a discolouration, a shadow in the wrong place, the same small wrongness three times in a row. The question that followed, what is that, was how the real problems were found.
_
_
Seeing is only the beginning of it.
Walk into a room for the second time and you are not shown the room; you are shown what has changed.
That is not a courtesy. It is a survey. Without consulting you, the change is checked against everything you have ever seen or heard or smelled, and a verdict returned on one question only: whether you are safe. The verdict arrives before you know it was asked, and everything else waits on it. If it comes back no, your intelligence is taken off you. If it comes back yes, you are free to wonder what you are looking at.
Other animals see better than we do, and hear and smell better. What we do that they do not is assemble all of it at once and then, having established that we are not about to be eaten, ask what it means. We have never been the best-equipped animal in the room. The asking is what we brought.
_
None of this is a gift handed to a special few. It is standard issue, and most of us mislay it.
I could once sit with a map before a journey, choose a route, and carry it in my head. I cannot do that now. My children were issued the same equipment but have rarely had occasion to train it. That was an errand rather than a faculty — work the mind was doing only because there was nothing else to do it — and I do not much mourn it.
What is on offer now is different in kind. It is not the route being taken off our hands; it is the reading. The machine will pattern-match faster than any of us, and it will be right most of the time. A thing that is right most of the time is very hard to argue with, and after a while you stop preparing to.
That the machines see more than we do is not the new part. Nothing ever needed us for the seeing.
It is the asking that is now optional.
_
The risk is not that the machines will notice for us. It is that a generation grows up with no occasion to sit in front of something unexplained and stay uncomfortable in front of it for a while. That they get no practice at the small insubordination of asking whether the answer they have been handed is actually right.
We are not going to lose the faculty. We will stop exercising it, until we can no longer feel that we ever had it.
_
_
I was seven or eight, at the kitchen table on a Friday evening, refusing my dinner. I know it was a Friday because we were eating fish, and I remember the fish because I hated it: a purée of tinned salmon, greens dried before packing and revived after, and a scoop of Smash. Pushing it round the plate, I fell to studying the wooden leg of the table, and wondered whether somewhere among its atoms there might be one that was a solar system like ours, with another small boy on one of its electrons at his own kitchen table, refusing his dinner.
Nobody asked me for that. It produced nothing. It has stayed with me for nearly sixty years.
_
I want the machines. I have wanted them since I stood in that clean room in my early twenties, watching one do what no hand could do.
But the person at the microscope was not there to be clever. She was there to look, and to be willing to be puzzled by what she saw, and to say so out loud in a room where an excellent machine had already given its answer.
A boy studies a table leg and gets nothing for it.
Nobody commissioned any of that. Forty years on, with a machine that can do very nearly everything else, those are the parts of the job I would not hand over.
_
22nd October 2026
_
© 2026 Paul Duggan
548 Market Street PMB 72296, San Francisco, CA 94104