The bottleneck has inverted
AI made producing documents almost free. The scarce thing now is judgement.
For as long as knowledge work has existed, the bottleneck was production. Writing the report, building the model, drafting the strategy — that was the hard, slow, expensive part. Checking it was comparatively cheap.
AI has inverted that. Production is now close to free. You can generate a competent first draft of almost anything in seconds. What hasn't got cheaper — what can't be compressed the same way — is verification: deciding whether the thing is actually right, whether it can be trusted, whether you're willing to put your name to it.
That matters more than it sounds. When production was the bottleneck, a signature on a document quietly did several jobs at once. It said: I wrote this, I reviewed it, I commissioned it, I stand behind it. Those four claims travelled together because the person who produced the work was usually the person accountable for it.
Pull production out and hand it to a machine, and the signature collapses. Authorship and accountability come apart. The interesting question is no longer "who wrote this?" — increasingly nobody did, not in the old sense. The question is "whose judgement is behind it?"
The value is moving from authorship of words to provenance of judgement.
It's moving not to the people who can produce the most, but to the people who can verify best — who know the domain well enough to spot what's subtly wrong, who carry the relationships and the credibility to be trusted, and who are willing to be accountable for a decision rather than just the words around it.
I think of it as human provenance. In a world where the words are cheap, the scarce and valuable thing is a chain of human judgement you can trust.
For anyone who works on the strength of their expertise, that's good news. The part of the job that was always the real value — knowing what good looks like, and being willing to stake your name on it — is precisely the part AI doesn't touch.