What to do when AI can do the thing you are paid for

Your pay is anchored to what replacing you costs, not to the value you produce. When the replacement gets cheaper, pay follows — and the return moves to whoever owns the thing doing the work.

You are good at something. You have been paid well for it. And you have watched a machine do a recognisable version of it, quickly, at a cost that does not resemble your salary.

The reassuring response is that it is not as good as you, which is often true and is not the point. The point is a mechanism, and it is worth understanding precisely rather than arguing about capability.

Your pay was never set by your value

As the wage post sets out, employers do not pay what your work is worth to them. They pay approximately what it would cost to replace you — and the gap between those two numbers is where their profit lives. That is not exploitation; it is how the arrangement has always worked, on both sides.

So the question is not whether a machine is as good as you. It is whether the cost of getting acceptable output has fallen. If it has, replacement cost has fallen, and pay follows replacement cost with a lag.

This also explains something otherwise puzzling: why pay can stagnate while your output stays excellent. Your value did not change. The alternative got cheaper.

Where the money goes instead

It does not vanish. The work still happens and someone still pays for it — the payment simply arrives somewhere else.

When a task moves from someone you hire to something you buy, the return moves from the person doing it to whoever owns the thing. That is the whole thesis of this site arriving at your desk rather than in an aggregate statistic.

Which tells you where to stand. Not “learn to use the tool” — everyone learns the tool, and a universally held skill commands no premium. The question is how to be on the ownership side of the transaction rather than the labour side.

What actually holds value

Three things resist this, and they are worth stating specifically rather than reassuringly.

Accountability. Someone has to be responsible when it is wrong. A tool cannot be sued, fired or trusted. Work where the buyer is purchasing certainty about the outcome rather than the output itself keeps its price — which is why the agency shift is from execution to judgement.

Deciding what should be done. Cheaper production is worth nothing to someone who does not know what to produce. The bottleneck moves upstream, and upstream is where the pay goes.

Context nobody wrote down. Knowing this client, this system, this industry’s undocumented reality. Not a permanent moat, and a real one for a long while.

All three are the same shift: away from producing the artefact, toward being the person who decides and answers for it.

The three responses

Move up within the wage channel. Toward decisions, accountability, relationships. Real and worth doing — and still the wage channel, capped by hours, priced by replacement cost, subject to the same pressure later.

Own the tools’ output. Use the cheaper production to make something you own rather than to do your job faster. Doing your job faster returns the gain to your employer; that is what employment is. The same productivity applied to something you own returns it to you.

This is the most accessible response for most people, and the one that actually changes which channel you are in.

Move to what compresses slowest. Physical, local, in-person work is under less pressure than cognitive work, which inverts twenty years of career advice. Plumbing is a more defensible position than copywriting right now, and that is a genuinely strange sentence to write.

What not to do

Do not compete on cost with the thing. You will lose, and the attempt lowers your rate while you find out.

Do not assume your case is exempt. Everyone believes their work needs judgement a machine lacks. Some are right. The belief is not evidence.

And do not wait for certainty. The mechanism is visible now; the timing is not, and it will not announce itself. The cheapest time to build a second channel is while the first still pays well — which is now, and which is exactly when nobody feels urgency.

The practical version

Keep the job. It funds everything, and panicking out of it is the worst available move.

Ask what you are actually paid for. Producing output, or deciding and being accountable? If the first, that is the pressure point, and it is knowable today.

Convert productivity into ownership. Whatever the tools let you do faster, spend some of the recovered time on something that accumulates. Start small, while employed.

And watch replacement cost, not capability. The question is never “is it as good as me.” It is “what does acceptable output cost now.” That number is the one your pay tracks.