There is a line about AI that has been quoted at me so often over the last three years that I now hear it in halves.
It came from the economist Richard Baldwin, on stage at the World Economic Forum Growth Summit in Geneva on 3 May 2023. What he said was: "AI won't take your job, it's somebody using AI that will take your job."
The half that travels is the first five words. The half that does the work is everything after the comma.
I have watched that sentence deployed as comfort. Someone in a meeting asks the question everybody in the room is already thinking, someone more senior offers the opening clause, the mood settles, the agenda moves on. Which is a strange use of it, because Baldwin was not offering comfort. He was pointing at a person. That person is not a machine, has a name, and is not you.
The parentage of that line is messier than a clean quotation makes it look, which turns out to be a small lesson in the same subject.
Baldwin's version is the well-sourced one. A stage, a date, an audience, a transcript. But an earlier and differently worded post had been circulating since December 2022, when the technologist Lucas Rizzotto wrote that AI would not replace humans, but that the people who worked alongside it would replace the people who refused to. In April 2024 he asked publicly whether he had accidentally coined the famous line, and said he would like to confirm it if it were true. He was asking, not claiming. Nobody has established the connection either way, and as far as I can tell nobody has seriously tried.
So the accurate summary is that the best-known form is properly attributed, an earlier near-relative exists, and the relationship between the two is unproven. I could tidy that into a cleaner story. I would rather leave it true.
Working that out took an evening and changed nothing about the argument. I went and did it anyway, and the reason I did it is the whole of what follows.
The best small-business marketing I have seen in the last few years belongs to an electrician.
Not an agency. Not a founder with a deck and a runway. A man who still spends his days in ceiling cavities and switchboards, and who has built a body of work around his trade that most marketing departments I have dealt with could not match on ten times the budget.
I met him the way you meet tradesmen, which is by needing one. Somewhere in the conversation it came out that he had built the whole thing himself, over years, in the evenings, and that it had been terrible at the start.
He is not technical in the sense my industry uses that word. He arrived with no advantage I could identify. What he has instead is a habit I have found rare: when he runs into something he does not understand, he treats it as a thing to be learnt rather than a thing to be avoided or handed to somebody else.
The order, as he told it, went like this. He learnt what a search engine actually rewards, which took him a while and cost him some wasted years. Then he worked out that his photographs were letting him down, and then he learnt enough about light to fix them. Then he learnt to write the way he talks, which took longer than everything else combined. When the AI tools turned up he did not announce anything or change his positioning. He added them to the list of things he was working out, the way he had added every other item on it.
Every one of those steps was available for purchase. He is not short of work and could comfortably have paid someone. He took them himself, because he wanted to know how the thing worked.
I have spent thirty years around people who had every advantage he lacked. Degrees, budgets, vendor support, and a good deal more raw ability than that electrician has. A fair number of them stopped learning at some point and did not notice it happening.
It rarely arrives as a decision. It arrives as getting good at something. You become the person who knows the thing, the organisation rewards you for knowing the thing, and gradually knowing it stops feeling like a position you are standing in and starts feeling like who you are. Then the thing changes, and you experience the change as an insult.
I have been that person. Long enough to recognise the feeling from the inside, and recently enough that I do not get to be smug about it. There is a specific flavour of irritation that shows up when a technology you were expert in stops mattering, and it disguises itself as a technical objection every single time. It is very convincing from where you are standing. It is transparent from anywhere else.
The electrician has none of that exposure, because he never let any one piece of knowledge carry his weight. He is not attached to any of it. If something better turned up tomorrow he would drop the lot and be mildly interested rather than offended.
Which is why I think the tool is the wrong object to be watching.
A tool that anyone can buy cannot stay an advantage for long. The moment it is good it is also cheap, and everyone in your market has it by Thursday. Every technology I have ever been paid well to understand has run the same arc: scarce, then valuable, then ordinary, then invisible. I have been around that loop enough times to stop mistaking the top of it for a career.
What stays scarce is the person who picks the tool up, works out what it is actually doing, notices where it is wrong, and puts it down again when it stops earning its place. You cannot buy that in. Nobody can take it off you either. It is closer to a temperament than a skill, and the only way I know to build one is to keep deliberately being bad at things.
That last part is the expensive bit. No point dressing it up. Being a beginner is unpleasant. It gets more unpleasant the more senior you are, because you have further to fall and an audience that has quietly stopped expecting it of you.
Read Baldwin's sentence again with the second half restored.
The somebody using AI who takes your job is not distinguished by owning better software. He has the same software you do. He is distinguished by having stayed curious long enough to work out what it is for, while you were being reassured by the first five words.
Everybody uses these tools now. That settles nothing, and it is not what separated the electrician from the people I watched quietly stop.
Measure yourself on something else. When did you last set out to learn something you were properly bad at, with nobody requiring it of you and no advantage guaranteed at the end of it? And if you have to think hard about that answer, what exactly do you believe has been protecting you all this time?