Preface

After I graduated, I worked as a contractor on Google Maps. One of my jobs was to find house numbers in Street View images and type in the numbers I could see. I didn't think much of it at the time. Years later, when I started learning how AI works, I realized what that job was: data labeling, the human work of telling a machine what the right answer looks like.

That realization is a good summary of this book. Behind every AI system there are people: people who collected the data, labeled it, designed the system, decided what it should do, and remain responsible for what it does. You cannot understand AI well without first understanding those people and the intelligence they were trying to capture.

By the time ChatGPT brought conversational AI into everyday use, I had spent more than seven years as a software engineer. I watched coding assistants learn to do more and more of the work I was paid to do, and I found myself a beginner again. There was so much AI material available that it was hard to know where to begin and which sources to trust. What I needed first was a clear foundation: what AI is, how it learns, why it can sound confident and still be wrong, and what stays with people.

This book is that foundation. I wrote it for the reader I was at the start: curious, willing to work, and in need of a map.

— Osman Mohammed