Human work becomes relatively expensive as AI trends to free
As AI-generated work trends toward zero marginal cost, the relative price of human involvement rises; the value delivered by humans must visibly exceed the AI alternative for the premium to hold.
The AI disruption conversation tends to focus on whether a given task can be automated. The corollary receives less attention: for tasks where AI can perform the same work at near-zero marginal cost, the price of having a human do it becomes, relatively, high. It does not matter that the human’s wage has not changed in nominal terms. What matters is the spread between the AI price and the human price, and that spread is widening.
The mechanism
Two things set this in motion simultaneously. The AI-generated version of the work keeps getting cheaper: compute costs fall, models improve, and vendors absorb more of the workflow. The human-generated version of the work does not. Salaries, overhead and oversight costs stay where they are. The result is that for any task both can perform, the AI version is a smaller fraction of the human version every year.
The pressure shows up in buyer behaviour before it shows up in pricing. Clients come to expect that anything that looks routine should be fast and cheap. When it is not, they notice. A price that was unremarkable a year ago begins to feel like a premium. Once it reads as a premium, buyers expect the extra to be justified.
What the premium has to carry
Human involvement survives commercially when it carries something AI cannot deliver — privileged context, accountability, trust, judgement under ambiguity. The value of that layer has to be visible to the buyer, not just present. An expert who cannot articulate what they add beyond the AI alternative is in a weak negotiating position even when they are, in fact, adding a great deal. The shape of professional communication shifts from “here is the work” to “here is what I did that the AI could not have done”.
This is the commercial side of what The relationship is the product argues for. When the work itself trends toward free, the packaging — relationship depth, continuity, demonstrable judgement — is what supports the fee.
What follows for firms
The strategic question for any firm whose billing depends on human hours is: what fraction of those hours is doing work an AI will shortly do at marginal cost? For that fraction, the price trajectory is downward and the competitive pressure is rising. For the remainder, the price trajectory is upward, but only to the extent that the buyer recognises the human contribution as different in kind. Firms that invest in making the human contribution visible will hold premium pricing for longer than firms that do not.