Knowledge work pools wherever measurement is easy. That's why so few people work on the bigger problems—the ones with names but no numbers. Every billable hour has to be justified to someone, and the justification runs through one question: How will we know it worked? Work that's connected to a metric—sign-ups, load time, retention, click-throughs—answers that question instantly. Work without one loses the argument for resources.
Journalism has already been through this. The moment pageviews made one kind of value countable, the work began to focus on it; the value that couldn't be counted (the beat and the sources cultivated over years) starved. The unmeasurable only continued to be done by the tenured who could afford it, and the stubborn who did it anyway.
AI thrives when working on the measured layer. And for anyone whose career sits on it, that reads as a threat—and partly, it is one. In the pageview era, people still competed against one another on the count. Now the machine does the measured work for everyone, and there's very little left on that layer to compete on.
Take any firm that sells expert time. Once the machine is doing the billable work, your salary is paying for work with no number attached. Most firms will cut it and call it efficiency. The ones that keep paying for the non-countable work are doing it for the least sentimental reason available: It's where an advantage can still live, if someone can make the case for it.
The people who already know how to make that case have been doing it for years: journalists, researchers, designers, and anyone who has had to get work past a budget meeting without a number attached to it. None of this makes the numbers matter less; nobody reports to a market or a board without them. But the path to a good number is about to get less legible than it has ever been, and whatever advantage there is now sits in the less legible layer.
