Ongoing research on the pressure to use AI, written for expertise-led business owners who are being told they are behind, covering what that pressure actually does to people and why the same pressure produces such different responses.
I was on Reddit, reading how people talk about using AI.
This started as marketing research. I wanted to know how the people I work with describe it in their own words, in the places where nobody is selling them anything and nobody is watching.
I did not expect to recognise it.
Every week I talk to expertise-led business owners living under the pressure to use AI. Sometimes it comes from a client. Sometimes from a competitor who will not shut up on LinkedIn. Sometimes it comes from inside their own head at two in the morning. The message is the same either way. You are behind. Fix it.
On Reddit it was that same message with the politeness stripped out.
What interests me is not whether that message is true. It is what people do with it.
Because they do very different things. Some take the pressure inside and reshape themselves around it, chasing whatever the market seems to be rewarding this quarter. Some step back and refuse, often knowing exactly what that refusal costs them. And some do the harder thing. They work out what is actually negotiable and what is not, and build from there. Still getting paid. Still recognisably themselves.
I have seen that shape before.
My doctoral research was about people under a completely different pressure, the pressure to turn yourself into a marketable product. Thirty interviews to start, then I followed the theory where it went. What came out of it were those same three responses, and I still think that theory is right.
So when this started showing up in my client conversations, I recognised it. That is the useful part and it is also the dangerous part. Recognising something is not the same as understanding it, and the differences between then and now are not small. Different people, different stakes, different thing being asked of them. The temptation is to point at my old work and say I told you so. That would be lazy and it would probably be wrong.
What I did not expect is what keeps showing up underneath all three responses now.
The gap between what people say they are doing with AI and what they are actually doing.
Performed adoption and real adoption are not the same thing, and the distance between them is where most of the interesting material is. Someone describes their AI workflow to me and then, forty minutes later, describes what they actually do on a Tuesday. Those are frequently two different accounts and almost nobody notices they gave me both.
It runs in both directions, which is the part I keep turning over. Some people claim far more than they do. Others quietly use AI for things they would never admit to using it for.
So I started collecting incidents. In the method that is a technical term. In practice it means moments. Things people said to me on calls, things they posted at midnight in forums where they thought nobody serious was reading, things I watched happen in rooms I was in. Each one gets coded on its own terms before I compare it to anything else, which is slow, and is the only way I know to find out something I did not already believe.
So I am not starting from nothing here. I have a theory in hand and I am pointing it somewhere it has never been tested. That is not the same as forcing it onto new material. If it holds, it holds. If what I keep hearing refuses to fit, the theory changes, and honestly that is the more interesting outcome.
I have a theory. I don't know yet whether it survives this.
I am writing about this as I go rather than disappearing for eighteen months and coming back with a finished argument nobody watched me build. There is a version where I wait until it is airtight and publish with footnotes. I have written that kind of thing before. It is a worse way to think, because the moment you commit to a conclusion in private you start defending it instead of testing it.
If you want to see what it looks like when someone gets to the other side of this, Pamela's story is one I point to often. She knew exactly what she wanted AI to do and could not get it to do it, which is its own particular kind of pressure. And if you are trying to work out where you actually stand before anyone puts more pressure on you, the AI Reality Check is ten questions built from this research.
This is live research. The incidents that don't fit are worth more to me than the ones that do, so if yours is strange, even better. If you have felt that pressure yourself, I want to hear about it. That is not a rhetorical invitation.