Every digital product you know added the word "AI" to its description sometime in the past two years. Some did it for good reason. Many others just changed the label.

A CRM running on conventional business rules for eight years is now "AI-powered." Tools that have been around since 2019 declare themselves "AI native." Native implies they were born that way. They weren't. They're converts.

In 1970, George Akerlof explained why markets collapse when buyers can't tell good products from bad. He called the bad ones "lemons" — used cars that look fine but are broken inside. When quality is invisible, buyers offer average prices, good sellers walk away, and only the lemons remain.

Michael Spence showed that costly signals can fix this — a university degree works because it's hard to fake. But when signals become cheap, everyone emits them and they stop meaning anything.

That's exactly where we are. The cost of "having AI" collapsed with language model APIs. Now every product signals it. The label no longer separates quality from noise.

This has happened before. ".com" in the 90s. "Cloud" in the 2010s. "Mobile-first." Each wave follows the same pattern: genuine innovation → valuable signal → signal gets cheap → everyone signals → signal becomes noise.

The economists mapped this territory sixty years ago. What they couldn't explain is why each new technology repeats the same sequence — as if the market had no memory.