Mark Zuckerberg thinks AI will start outperforming humans in the next decade (original) (raw)

On yesterday's investor call, Facebook founder and CEO Mark Zuckerberg was asked how the machine learning technology behind its recent introduction of bots to Messenger would manifest itself in the future. "So the biggest thing that we're focused on with artificial intelligence is building computer services that have better perception than people," he replied. "So the basic human senses like seeing, hearing, language, core things that we do. I think it's possible to get to the point in the next five to 10 years where we have computer systems that are better than people at each of those things."

There are a couple of striking things about this statement: it imagines computer systems with abilities that are central to our existence as biological creatures, the essential skills that we rely on to understand and interact with the world. And it then assumes that a wide variety of jobs currently done by people will no longer require expensive human labor, because artificial intelligence will be superior.

"That doesn't mean that the computers will be thinking"

Zuckerberg does draw an important line in the sand. "That doesn't mean that the computers will be thinking or be generally better," he says, faint but important praise for us meatbags. Thinking in this context means something more complex than understanding how to carry on a conversation or tell a cat from a dog. It's about being able to do a wide variety of different things well, synthesizing disparate topics, and learning on your own in an unpredictable and unsupervised environment. "Generally better" is being used in the sense that we are generalists, capable of handling a wide universe of tasks.

We can build a machine that beats the best human at the game of Go, but that same machine can't stop in the middle and write a poem about about its style of play. It can't learn a much simpler game, like checkers, without lots of human help first. Yan Lecun, the director of AI research at Facebook, put it this way:

"We know now that we don't need any big new breakthroughs to get to true AI"

That is completely, utterly, ridiculously wrong. As I've said in previous statements: most of human and animal learning is unsupervised learning. If intelligence was a cake, unsupervised learning would be the cake, supervised learning would be the icing on the cake, and reinforcement learning would be the cherry on the cake.

We know how to make the icing and the cherry, but we don't know how to make the cake. We need to solve the unsupervised learning problem before we can even think of getting to true AI. And that's just an obstacle we know about. What about all the ones we don't know about?

The rub, of course, is that lots and lots of jobs don't require a renaissance man or woman. Here's a chart from The Wall Street Journal based on research done by two economics professors. It shows the trajectory of jobs, both manual and cognitive, over the last four decades. Manual labor hasn't ticked up much, even when the work involved requires you to be adaptable. Non-routine cognitive work has steadily increased. But routine work, whether it requires your mind or your body, has fallen.

Zuckerberg sees bots stepping in for customer service reps and personal assistants. "One way that I think you're going to see bots work, between people who are actually driving the businesses directly will need to in some way train or answer questions for people, but we can build artificial intelligence that can learn from people how to automate a lot of that and make that experience a lot faster for people who want to interact with businesses and public figures."

Basic income is one solution

Facebook wasn't the only one focusing on artificial intelligence during its earning's call. Google CEO Sundar Pichai also made it one of the central themes of his remarks. How should we adapt to this new reality, if Zuckerberg's predications are correct? One possibility is universal basic income, currently a hot topic around the dinner table in Silicon Valley. The other is adapting our educational system to focus on fostering future generations equipped to pursue the complex cognitive tasks that will be the heart of our available labor pool in a world of advanced AI.

Or you could look at the spectacularly underwhelming launch of recent smart bots from Microsoft and Facebook — they turned out to be racist and not very smart — and laugh all these predictions off as a other AI hype cycle that will quickly fizzle out.

Facebook’s big bot bet