How the World Got Wired

Chapter 21 — 2012 and After

The year 2012 was, for Geoffrey Hinton, a strange year. It was the year of two completely unrelated decisions, both made for ordinary human reasons, that ended up reshaping the world.

I am going to tell you both of them in this chapter. The first one ended in disappointment. The second one ended in the breakthrough that began the artificial intelligence revolution you are growing up inside.

A Father, Doing Math

By 2012, Geoffrey Hinton was 65 years old. He had been a professor at the University of Toronto for twenty-five years. He had a respectable academic salary, modest by American standards. He had been working on neural networks, with patient Canadian funding, for almost his entire career. He had not become rich.

He had a personal situation, which he has spoken about publicly in many interviews, that I will describe carefully and briefly, because it matters for this story and because Hinton has chosen to share it himself.

Hinton’s son needed, and would continue to need, professional care for the long term. The kind of care that costs a great deal of money, every year, for many years. The kind of math any parent in that situation has to do involves sitting down with a calculator and projecting forward — how much will this cost over the next ten years, the next twenty years, after I am gone — and arriving at a number that is, frankly, terrifying for most people.

A Canadian university professor’s salary, however carefully saved, was not going to cover it.

This is a quiet, hidden engine that runs underneath many decisions in the world that get reported as if they were about something else. Many famous people, in many famous moments, are doing private math about their families and their futures, and arriving at conclusions the public never sees. Hinton, in 2012, was one of them.

He decided he had to make some money.

The Coursera Course

His first attempt was charmingly unglamorous.

In late 2012, the online education platform Coursera was new and growing fast. Coursera let university professors record video lectures and offer them, for free or for small fees, to anyone in the world with an internet connection. Some professors who taught popular courses had attracted hundreds of thousands of students. A few had begun to earn meaningful amounts of money from premium features and certificates.

Hinton, who knew more about neural networks than almost anyone alive, decided to teach a course. He set up a camera in his office. He spent weeks recording video lectures. The course was called Neural Networks for Machine Learning. He released it on Coursera in late 2012.

The course was, scientifically, magnificent. Thousands of students enrolled. Many of them, watching Hinton patiently explain ideas he had spent his life developing, became neural network researchers themselves. Several of them, over the following decade, went on to do important work in the field. The course is still considered, by people who study artificial intelligence, one of the best introductions ever made to the subject.

In strictly financial terms — the terms that mattered for his son’s care — it made him very little.

This is the part of the story I want you to remember when, later in life, you hear adults tell you that talented people will always be rewarded for their talent. Sometimes they are. Sometimes they aren’t. The same Geoffrey Hinton who, six months later, would be at the center of a bidding war between the largest companies on Earth was, in 2012, the kind of teacher whose Coursera course made roughly the same income as the Coursera courses of less famous professors.

The way Hinton would eventually solve his family’s financial problems was not through teaching. It was through something else.

That something else was happening, in parallel, in his lab.

All 24 chapters of How the World Got Wired · Library · All courses