Personal Essays · 8 min read ·
The work we were never meant to do
The oldest tool in a white-collar day is forty-seven years old, and taking the AI off the desk leaves about fifteen minutes of that day standing. What is inside the gap is not something lost. It is forty years of borrowed work.
The oldest tool that makes up a white-collar day is forty-seven years old. The first spreadsheet went on sale on 17 October 1979, the first presentation software on 20 April 1987. The skeleton of your eight hours is not a chapter in the history of human work. It is a footnote.
Here is the same span from one country. In 1991, forty-six of every hundred working people in Turkey worked the land. In 2025, fourteen. One generation left the field, moved to a city, and now watches a grandchild sit at a desk, and that generation argued for thirty years about whether it had been worth it. You will struggle to find anyone asking today. The question did not get answered so much as run out of people who had seen both sides of it.
So I asked a smaller version of it about myself. If the data centres were bombed and the AI came off my desk tomorrow morning, how much of what sits in my calendar could I do end to end? I made the list. On a competitor analysis I do not gather the data, write the summary or build the table; I decide which question to ask. On a deck I do not build the skeleton or write the sentences; I say which sentence is wrong. Both of those are decisions, and together they take about fifteen minutes.
So what was filling the other seven and a half hours?
Looking into that gap I did not find something I had lost. I found something I had been borrowing for forty years. It has happened before, and on roughly this schedule: electric motors were available from the 1880s and factory productivity did not move until the 1920s; fission was discovered in 1938 and produced its first usable electricity on 20 December 1951. Every time, the tool arrives early and the organisation changes late, and the forty years in between get filed afterwards as a transition period.
Whether a transition was good or bad is a question only for the people standing inside it. We are standing inside one, and the claim of this post is that we were never meant to be doing that work in the first place.
The fear is not about the job
I had assumed the fear was about losing work, and it is not. Someone afraid of losing their job knows exactly what to do and has nowhere to do it, while mine ran the other way round: the desk is still there, and what I cannot remember is what I used to put on it. It felt less like a dismissal and more like the moment you notice you have been a passenger in a car you were sure you were driving.
That is what the list did. Missing a tool and not knowing what is underneath it are different feelings, and only one of them survives being written down.
The void is forty years of borrowed work
Ask yourself how many of the things you did in the last twenty-four hours you were ever actually obliged to do.
I am not asking about work that takes daily skill, gets done with your hands and shows its result at the end of the shift. A cook's day is not the subject here and neither is a nurse's, because nobody can describe either one as something that need not have happened. I am asking about white-collar work, and specifically about the crowd that feels busy and produces nothing you can hold in your hand. I am in that crowd.
Calculating something in Excel. Building a deck that gets opened once, in the meeting it was built for, and never again. Running a projection. Producing content, tracking how that content performed, reporting on it, then shouting at whoever is below you about it.
All of it is new. To see how new, look at when the tools that make up your working day were actually invented:
Forty-seven and thirty-nine. That is the age of the oldest piece of an office day and the age of its most familiar one, and both are younger than most of the people who open them every morning. Your job title is probably younger than your parents' marriage.
Now look at that from two hundred years out. In ten thousand years of human work, we are defending a set of tasks invented inside a forty-year bracket as though they were part of what a person is. Standing at that distance I reached one conclusion, and it is the one this whole post rests on: we were never supposed to be doing any of them ourselves.
Millions of people already knew this
The thought that pushed me was this. For thirty or forty years, millions of people have been going to work at jobs they should never have had to do.
I am not the first to notice. David Graeber described these in 2013 as jobs whose holders "secretly believe" do not need to be performed, and two years later YouGov put a number on it: 37% of working British adults said their job was not making a meaningful contribution to the world. Half said theirs was, and 13% could not decide. The feeling is neither universal nor marginal. It is a third of the room, and it does not move.
This is where the fear changed sides.
When I looked at the void again a week later it had already started filling on its own, because imagination puts things into any space you leave empty for long enough. So I asked what I was interested in before AI, and the answer came back quickly: making ideas, marketing them, putting people together. Those were good days and I remembered them clearly. Take out the work I did only because it was easy, the work that lived off this transition, and what remains is exactly that list, which is also exactly what I was doing before any of this started.
The reason that reads as a relief rather than a loss is not optimism, it is accounting. Losing a job is only a loss if the job should have existed in the first place, and to call the closing of a forty-year bracket a loss you have to believe the bracket was the main text rather than a footnote.
Forty years of borrowed work is like renting a flat for long enough that you start calling the furniture yours, and then getting upset when the landlord takes the sofa back.
The void was not the size of what I lost. It was the size of what I had been borrowing.
We will call these forty years a transition period
Two hundred years from now, these forty years get filed as a transition period.
I passed over the reason at the top, so here it is. The forty-year lag has nothing to do with how hard the tool is to learn.
Paul David's 1990 paper tells the story of electricity this way. The first thing factories did with electricity was use it to turn the old steam shaft. The belts stayed, the pulleys stayed, the line shaft in the ceiling stayed, and only the thing at the far end of them changed. The gain arrived forty years later, when the belts came down and every machine got a motor of its own. Those forty years went into rebuilding the building, not into learning the tool.
Fission moved faster and was still slow: thirteen years from the discovery to four light bulbs in an Idaho desert, sixteen to the first confident promise about what it would mean for everybody else.
Three years later, on 16 September 1954, Lewis Strauss of the Atomic Energy Commission told a room full of science writers that our children would have electricity in their homes "too cheap to meter". They did not. The sentence survives today as the standard example of an industry promising more than it delivered.
I am not writing that to make Strauss look foolish. The man standing in the middle of a transition turning out to be the worst reader of its far side looks less like a personal failure than a rule, and you and I are currently standing exactly where he was standing.
We are the last generation that can tell the difference
Here is the question that actually matters. How many people ever live through the good-or-bad argument?
Only the ones inside the transition. Us.
Think of it as the last generation in a village that still speaks two languages. They are the only people who can tell you what the older one was good at, because everyone after them speaks only the new one and has nothing to miss.
The first of the three rules Douglas Adams wrote down on 29 August 1999 is that everything already in the world when you are born is "just normal". AI will be normal for my grandchild in the way that running water is normal for me, and with no "before" to hold it against, the question of whether it turned out good or bad will simply not occur to them.
So the best answer to "will AI be good or bad" is still "both". But the answer is not the important part; the important part is that ours is the only generation in a position to give one, because we lived the before and we are the ones building the after.
Do not assume history will grade this on your behalf. A grade needs somebody capable of comparing, and that person exists only during the transition. The people who come after you will not ask whether the tool was used well, they will simply inherit how it was used.
Being early here is not a grand thing, and it is worth saying what it actually is: you are among the last people who can keep a record of which task you handed over and why, and which one you kept and why. In forty years that record gets read by somebody with nothing to compare it against. What you leave them is not an argument. It is a habit.