Introduction

My garage door started opening itself and refusing to close. I drove halfway to New York City one afternoon before I realized, turned around, came back, closed it by hand. The local repair guy quoted $1,200 to replace the sensors. Before agreeing, I took four photos of the assembly and uploaded them to ChatGPT. It gave me four debug steps. By step three the door was working.

Step three was clean the lens.

That's the version of the AI story everyone tells. The version you've heard a hundred times. The expert charged too much; the chatbot solved it in two minutes; the future is here and it's almost free.

This book is about the other version. The one nobody writes about because it's harder to celebrate and harder to sell. Not whether AI works, it does, increasingly, well enough to be unsettling, but what it costs, in the places the productivity pitch doesn't mention.

We're still in the shallow end. AI isn't a tractor. Tractors took a hundred years to reach every farm; they needed a showroom, a loan, and a road to get to you. AI arrived on every phone in every pocket at roughly the same moment. There's no slow on-ramp. There's no transitional generation.

The adoption curve is vertical, and the adaptation curve, the human, organizational, cultural work of figuring out what this actually changes about us, is barely a line yet.

Some people are running toward it. Some are bracing against it. The companies selling it are doing both, depending on the audience. I'm in the running-toward camp, I use these tools every day, I build with them, I'm writing this book on one. But running toward something isn't the same as understanding it, and the closer I've gotten the more I've noticed what the decks leave out.

The lens-cleaning version of AI is real. So is the other version. This book is about the other version.