Facts are friends
Start here
What's actually changing - and why it is easy to miss
The quiet kind of change
I have shopped at the same Sainsbury's in Warwick for many years. A decade ago it had a full line of checkout staff, others covering breaks, someone on the customer service desk. Today it is mostly self-service machines, one or two people overseeing a dozen kiosks. Smart shelves and handheld scanners have changed the restocking role. Even the night shifts are thinner than they were.
Nobody voted on this. There was no announcement, no headline, no public debate. It arrived a little at a time, over ten years, until one day the person you used to say hello to was no longer there.
That is what most AI-driven change actually looks like from the outside. Not a dramatic confrontation. A quiet thinning that is easy to explain away, right up until it is not.
It is not only the obviously mechanical jobs, either. Translators have watched their own carefully rendered sentences become the raw material that trained the machines now competing with them. The "clever work" that happens over the water cooler - the relational, tacit knowledge people assumed only a person in the room could pick up - turns out to be exactly the kind of thing a camera and a good model can learn by watching. Nothing about being human-shaped work has made it safe.
Faster than the reassuring version, slower than the alarming one
The honest answer to "how fast is this happening" is unsatisfying: faster than most people in authority are willing to admit, and slower than the most excitable technologists would have you believe.
The capability - what the technology can actually do - is moving at a pace with no real historical precedent. The adoption of that capability into ordinary working life is considerably slower, for entirely ordinary reasons: infrastructure takes time to build, trust takes time to earn, regulation takes time to catch up, and human habits take longest of all.
Electric vehicles were genuinely good by 2012. Mass adoption still took years, not because the cars were lacking, but because charging infrastructure needed planning permission, grid upgrades and a dozen organisations to coordinate. Contactless payment existed for years before it mattered, and then COVID arrived and it became essential infrastructure within months. AI displacement, I think, is following the same shape: a long quiet build, then a tipping point that feels sudden even though it was not.
This kind of change is never a single event. It is a process, slower than its proponents predict and faster than its subjects prepare for.
What this means practically
We have time. Not unlimited time, and not as much as we would like, but time. The most consequential effects are not yet fully visible in the employment statistics, and that is not evidence the whole thing is overstated - it is evidence we are still inside the preparation window, the gap between the capability arriving and its full effects landing. That window is the most valuable thing available to us right now, and it will not stay open indefinitely.
Three questions worth sitting with
What has changed in your working life, your high street or your profession in the last ten years that happened without anyone quite deciding it should?
What change do you already know is coming - in your sector, your role, your community - that is currently being delayed by regulation, cost or simple institutional reluctance to look at it directly?
And who should you actually be talking to about this - not in six months, not after it becomes unavoidable, but now, while the choices in front of you still have room to move?
Once you have sat with those, the next question is a practical one: what, specifically, is worth doing about it.