Former Google CEO Eric Schmidt has a fantasy.
Ten or 20 years from now, a machine solves a science problem “to the point where the computer will get the Nobel Prize.”
“It may not be probable, but it’s possible,” Schmidt said during a wide-ranging podcast interview with MIT Sloan principal research scientist Andrew McAfee. “I’m most optimistic about the rate at which we can use these new approaches in machine learning and AI to solve problems that are really, really hard.”
In his conversation with McAfee, co-director of the MIT Initiative on the Digital Economy, Schmidt talked among other things, about machine learning worries and how he managed Google’s co-founders.
In the last quarter of 2017, Google brought in $27.2 billion in advertising revenue (or 85 percent of parent company Alphabet’s revenue) — a far cry from 17 years ago when Schmidt joined the startup tech company.
“When I first came in I was so convinced the ads didn't work, I hauled in the sales executive, who was Tim Armstrong at the time, and said to him, 'Prove to me these ads work,'” Schmidt said. “The way he proved it to me is he showed me the bank accounts, that people were actually paying for them. Today of course, that’s a $50 billion-plus business.”
When Google co-founder Sergey Brin had an idea — which would ultimately become Google’s AdSense for targeted ads — Schmidt gave him a $1 million cap to test it.
“He comes back and he’s spent $1.5 million,” Schmidt said. “Such disorder from a founder. Of course, that $1.5 million that he spent became a $10 billion business. It gives you a sense of the scale of how big these businesses could be.”
Schmidt joined Google in 2001, and he credits part of his hiring by Brin and co-founder Larry Page, to being a fellow fan of desert arts festival Burning Man.
“I would treat them as ‘the boys,’ the young men that I worked with; with respect but almost a fatherly way,” Schmidt said. “There was one day we were having some huge argument over strategy, and I realized these aren’t ‘the boys’ anymore, these are men who’ve been through some of the toughest business challenges in the world and resolved them.”
That was almost a decade ago, Schmidt said, and at that point, it became about how could he assist with their leadership.
“When you’re in the adult supervision department, there is a point where you should stop,” Schmidt said. “I was sufficiently humbled very early with Larry and Sergey, that I learned to not just dismiss their intuition. When you work with people that brilliant it’s very, very important to understand what is it they’re trying to achieve and how do I help them achieve it.”
Schmidt joined MIT earlier this year as a visiting fellow and adviser for the MIT Quest for Intelligence. The school-wide initiative studies human and machine intelligence.
While a computer winning a Nobel Prize might still be a decade or two away, Schmidt said in the next three to five years, he’s optimistic about machine learning and artificial intelligence solving tough problems. The thing he’s worried about, however, is industry being too far ahead for governance.
“I’m worried that we’ll screw that up,” Schmidt said. “That somehow the enormous engine of innovation will get slowed down because of legitimate thoughts which have not been fully planned out about the effects that they have on people.”