Three years ago a legal will cost $400 to draft. Today AI does it for $0.50. That productivity gain is real. It also doesn't show up in a single economic dataset.
SemiAnalysis calls it "Dark Output": AI is creating massive economic value that GDP, employment data, and national accounts cannot see. Like dark energy in physics, you can only detect it through its effects on other things.
The fingerprint: employment in AI-exposed sectors is falling while wages in those same sectors are rising. Not because anyone got a raise. Because AI displaced the cheapest workers first and the average moved up mechanically. The Fed sees rising wages and thinks the labor market is tight. In reality AI ate the bottom of the labor stack.
Incoming Fed Chair Kevin Warsh said it himself: "If you are looking at the data, you are backward looking. You are going to be late." He's telling you the instruments are broken.
The implications: the people calling AI a bubble are measuring the wrong side of the ledger. The Fed is making policy with broken instruments. And the tax base governments depend on is shrinking while actual economic activity is not. That gap has real implications for anyone holding assets that exist outside the measurement system entirely.
