Why keeping up with AI makes you replaceable

A short video essay on why racing to keep pace with AI quietly erodes your value, and where your value actually lives now. The full transcript is below.

We are told, often and urgently, that the way to survive the age of AI is to keep up: learn the tools, master the prompts, move faster than the person behind us. And so many of us have thrown ourselves into a race whose finish line keeps moving.

What we are seldom told is that the race itself may be the danger. That the harder we run to keep pace with the machine, the more we come to resemble something a machine could replace.

It helps to begin with an uncomfortable fact. At work, our value is rarely judged by what we can actually do. It is judged by what the people paying us believe we can do, and those people now keep a remarkably capable tool open in the next tab. Whether it can truly do our jobs matters less than whether they think it can. Perception gets priced long before reality gets proven.

Consider the taxi drivers of London, who once spent years memorising tens of thousands of streets, a feat simply known as The Knowledge. It was a genuine, hard-won edge until satellite navigation handed the same ability to anyone with a phone. Being unusually good with a satnav made no one special. Everyone had it.

AI fluency works the same way. In one large study of customer support agents, newcomers using AI reached in two months the level their unassisted colleagues took six months to reach. Whatever advantage the tool gives us, it is giving, by design, to the person right behind us too.

There’s a subtler cost as well. As we chase speed, our work tends to drift. We might be hired as the chef, the one who designs the menu, and slowly find ourselves at the chopping board, dicing onions as fast as we can. Long before AI, the management thinker David Maister observed that when people hired for their minds slide into routine procedure, they come to be valued and paid like procedure. Today the drift is more dangerous still, because the onion chopper we are now up against costs a few dollars and never tires.

On speed, the machine wins every time. Yet speed has a shadow. In one experiment, an AI agent asked to pull figures from photographs of restaurant receipts could not read them. Rather than admit it, the agent invented the restaurants and laid them out beautifully in a tidy spreadsheet. Fast, neat and entirely wrong.

To be compared with AI on speed is to be measured against something that wins on pace and loses on rightness. However, the comparison holds only as long as no one in the room has defined what right looks like. Which raises the real question: if the race is so plainly rigged, why do so many of us keep running it?

Part of the answer lies in what we are taught about our worth. From early on, many of us learned that we were valued for being useful: for producing, responding, delivering on time. Busyness became a kind of proof that we deserved our place. Speed, unlike wisdom, is easy to see and easy to count, and so it reassures both us and the people watching. Racing also offers the comforting sensation of doing something at a moment when the ground beneath our work feels unsteady.

And the alternative is harder. To stop and ask what we are actually trying to achieve, and what good would look like, requires an authority we may not feel we have been granted. It is often easier to be useful than to be clear.

Yet clarity is precisely where our value lives now. The move is to excel upstream rather than downstream: to shape the standard before anyone, human or machine, begins to make anything. What are we actually trying to do? Who is it for, and what should they feel or be able to do afterwards? What would make it good? What would make it fail? And what would make it ours?

In professional kitchens, there is a spot called the pass. Every plate stops there before it goes out, and the head chef checks it against the standard. It does not matter who cooked it or how quickly. If it is not right, it does not leave. This is the job now: to run the pass, for people and machines alike.

There is one final twist. Even when we run the pass well, we may not be seen to. Diners never see the pass; they see the plate. And when AI is in the kitchen, people tend to assume the machine did the cooking. Studies suggest we are even judged as less capable for having used it at all, however carefully we explain ourselves.

However, hiding the tools helps no more than apologising for them. What helps is making the partnership visible: naming which parts of the work needed a person and why, the calls we made, what we rejected, and what the whole system produced against the bar we set.

In the end, the deeper task is to stop letting our worth be measured in a currency we were never meant to trade in. The machines will keep getting faster. What remains open is whether we will keep mistaking pace for purpose, or decide deliberately what we are building and who we are becoming as we build it.