Someone will tell you this week that artificial intelligence only assists human work and will never replace it. The claim comes from executives and from writers who cover the technology for a living. It reassures, nothing more.
It is also too late for that reassurance to still be true. The replacement is already running. Companies make decisions this quarter, on an ordinary Tuesday, and some of those decisions remove a person from a task they used to be paid to do.
The larger question, whether human labour or software handles the next years of work, is close to tipping one way. Most people would rather not watch it tip, because everyone wants their own job to sit on the safe side of that line. But the choice isn’t really individual. Other people make this decision, in boardrooms and product roadmaps far from any one desk. Even that overstates how organised it is: there is no single group steering the outcome any more. The thing has taken on its own momentum, and nobody involved, including the people who built it, seems able to stop it now.
The money already committed to AI infrastructure moves the same way. Companies spent it before anyone downstream worked out what to build with it, and every smaller decision since depends on that spending having already happened. Nobody can call it back from a single desk.
None of that makes uncertainty dishonest. Nobody knows where this settles, and saying so plainly is the honest position. Dishonesty starts the moment someone keeps insisting that people simply cannot be replaced, as though the outcome were already decided in humanity’s favour.
Not every task a person can do turns obsolete on the same afternoon. I’m not compiling a list of future-proof jobs here, and I’m not trying to prove how essential the rest of us remain.
What I want is some sense of the scale: how many people end up without a job because AI has taken over the work they used to do.
I can’t give you a number, and I don’t think anyone can either. The reluctance sits on many desks at once, quietly and without coordination, closer to a shared habit than a plan anyone wrote down. Almost nobody wants to look straight at the number.
I know that reluctance from the inside.
For years I sold something without being sure it worked: a wristband and a story that was almost true. Each morning I put the glasses on, because they felt better than the question of what I was really doing.
When someone tells me today that AI only assists and never replaces, I hear the same tone I used on myself back then: someone trying to believe a thing because the alternative is too large to face. I don’t blame him for it: I did the same thing myself.
Taking the glasses off improves nothing about the view. It shows where you actually stand: the replacement is running, and the count of the people it displaces does not exist yet.