IEET Published by Patrick Tucker and The Futurist magazine. Article by Geordie Rose.
IEET Published by In the future, humanity is going to try to build things like David. We’re going to fail at first, because that’s the nature of big things. But just because you’re probably going to fail doesn’t mean you shouldn’t try to do it. If we at my company, D-Wave, listened to everybody that said that you couldn’t build a quantum computer 10 years ago, we never would have tried.
But we did try. And we succeeded. These two goals have a lot in common.
The original goal of AI was not to build things that make you click on ads more. That wasn’t the reason that AI got started. The reason AI got started is because we wanted to build the things that were like those science-fiction robots. We wanted to build machines that behaved more like us.
Over the years of failure, and determining how hard this problem was, people got disillusioned. Now, it’s not even talked about in polite company anymore, although it’s coming back a little bit. I never really cared what people think. I think that this is worth doing. And we should try to do it. … At D-Wave, we’re trying to build systems that use all different resources in a way that allows us to try a huge amount of things to try to build truly intelligent machines.
Part of the reason D-Wave was a success was that we took the attitude that we couldn’t know in advance what the right answer was. We didn’t know what the right design was for a quantum computer. So how do you get around that? You try thousands of ideas as fast as you can. And you evolve the solution.
So what we’re trying to do now is set up an infrastructure that will allow us to try tens of thousands of ideas, and narrow the solutions. We’re going to try to apply the things that worked well with building quantum computers to this even harder problem of building machines that are intelligent in the way that we think we are.
Today, our computers are little clockwork universes that we’ve created within a chip. And those little clockwork universes can do an awful lot of things very, very well. But they are not like the way nature actually is.
We know that nature isn’t clockwork, but that used to be a controversial point of view. The clockwork universe assumption, popular from the seventeenth century until the twentieth, was driven by an understanding of the world that was developed during the Enlightenment: that the universe has gears and clicks forward in a very deterministic and linear way. Therefore, if you know something at some point, you can always know what’s going to happen at another point.
Nature is fundamentally different from a clockwork universe. It’s far more complex. Once you acknowledge that, you have to acknowledge that it’s possible to build machines that are not like our computers—machines that are like nature. These machines can solve problems that you couldn’t otherwise solve. Quantum computers are one type of machine like that.
It’s the way that they do their computation that gives insight into something about the ontological nature of reality. What is actually out there? If we rely only our senses, we’ll never know. We see the world through lenses that are so thick, it’s amazing that we can know anything else about the universe other than what we see.
The amount of data that’s inundating the room you are sitting—the cell-phone signals, the Internet, the gamma rays, even just the photons that you can see in that tiny little spectrum that we can see with our eyes—it’s such a tiny part of the universe. It’s fascinating we can do physics at all. But quantum mechanics takes that concept to a whole other level.
You see, in quantum mechanics there’s a perfectly viable explanation for the way the world works, which is the idea that every time a decision is made, every time a potential becomes a reality, the whole universe forks and generates copies of itself.
Our way of computing at DWave is very different from the ways that people build computers today, architecturally. Our way is a lot more like a brain. It’s more like a neural network. Neurons, in this case, are molecular devices called qubits, or quantum bits. They’re like neurons, except they’re quantum.
The progression of quantum computing technology over the last eight or nine years has been exponential. And the number of these qubits—neurons on the chip—has been steadily doubling every year for almost nine years now. As you get this number larger and larger, you start pushing into territory that allows you to do things that you simply can’t do with conventional approaches to computing. The benefits aren’t just speed.
Nevertheless, there are a lot of things these computers can’t do well. Why aren’t there any New York Times bestsellers written by computers?
The problem of making a machine that thinks like us is a lot like the problem of making a quantum computer—it’s the sort of problem where there’s no good reason why you can’t do it. Doing it right requires a lot of money. And time. But that’s not a reason to not do it.
So perhaps in 10 or 15 years, it won’t be me telling you about this. It will be something that we created.
About the author: Geordie Rose, founder and chief technology officer of D-Wave Systems Inc., is the creator of the D-Wave One, the world’s first commercial quantum computer.
This article was adapted from his presentation at WorldFuture 2012, the annual conference of the World Future Society.