h+ Magazine

Off a busy suburban road and inside an ordinary office complex, the future of computing may be taking hold at the subatomic level.

At first glance, D-Wave Systems looks like any other well-appointed office, with an open reception area and conventional cubicles. But one glance at the wall beside the receptionist and you know the average IQ here is intimidatingly high — it’s literally covered in plaques from the U.S. patent office featuring 19th century lettering and incongruously describing patents for superconducting qubit-based microchips.

As I step past reception, I’m ushered into a boardroom with a wall full of whiteboards filled with equations. Founder and Chief Technology Officer Geordie Rose greets me. He resembles a cerebral Ray Liotta. As we sit down, he explains the ultimate goal of his company: to help achieve AI that’s human-level… or better.

Unlike a conventional computer that works in a binary process of rapidly switching circuits on and off, a quantum computer uses subatomic phenomena to create logical circuits called qubits that are on and off at the same time. Qubits can be a 1, a 0, or a quantum superposition of both. Put a pair of qubits together and they can be in a superposition of four states. With three qubits you can have eight states, and so on exponentially. D-Wave Systems intends to push the field further than ever before with the world’s first 128-qubit computer, running on superconducting circuits.

It all sounds very exotic, but quantum phenomena are used today in lasers and other technology we take for granted. D-Wave’s quantum computers are currently engaged in optimizing learning algorithms that can be transferred to conventional computers. Rose told me D-Wave recently worked with Google on the “Google Goggles” augmented reality mobile phone application by using their systems to “teach” a neural network how to recognize objects like automobiles, and then transferring those optimized algorithms to the mobile application.

In fact, machines can already perform almost any narrow task much better than humans. It’s the bringing together of those skills into a more general framework that Rose sees as the crucial step towards true AI. As Rose puts it, “Learning without ‘hand-holding’ is the last missing piece.” He also pointed out that AI research is moving at an unprecedented pace thanks to the drive for better mobile phone apps.

“Microsoft, Google, Apple and other companies all want to dominate the mobile space, and to do that you need compelling applications. The potential is extraordinary, like mobile devices that monitor your health, or talk back to you when you talk to them. All of that requires better AI, so massive amounts of money are going into that space.” Rose believes that D-Wave’s quantum computers will help steward this renaissance by acting as “tutors” for the myriad devices that will require optimal AI algorithms to compete in the marketplace. Rose: “I’m very excited by the possibility of building very effective unsupervised learning systems and contributing in a meaningful way to the creation of better-than-human level intelligence in machines. With a very fast solver, you can try many things in a short period of time and zero in on methods that work.”

The quantum computing industry isn’t mature yet. It’s currently on the other end of the hype cycle from the mobile space, with information about the field scattered and generally beyond the read of most non-experts. In Rose’s opinion, quantum computing isn’t powerful enough yet to be useful to a broad range of people, but “it’s conceivable that we could make it extremely powerful in the next three to five years.”

Looking forward 50–100 years, Rose has absolutely no doubt that machine intelligence will far outstrip that of humanity. “The existence of vast machine sentience is almost guaranteed to occur,” Rose told me. “You can ask your cell phone what it’s thinking about now, and the answer is that it isn’t. But in 50 years it will be. And it won’t be a companion of yours, you might be a companion of it.”

But will these smart machines lead, as most transhumanists believe, to the Singularity? Rose doesn’t think so. “I think the term ‘singularity’ is flawed, because it implies a single event. If there is a single event, we already had it when we invented the transistor. I get the feeling the early days are here already.” And while Rose believes in the feasibility of life extension, he thinks it too will be based on silicon, rather than tinkering with our current biological template. We will enter the future as machine intelligences. “Our bodies fail in a lot of different ways, and it’s likely we can’t extend those bodies past their ‘best-by’ date, but physics says what we are as humans isn’t dependent on a substrate. There’s no physical reason why it isn’t possible to replace that substrate with silicon.” He adds, “I question the belief that ending death is ‘too hard.’ There’s no reason we can’t find a way to break the cycle, and the way to defeat it is to remove the substrate.”

As I leave D-Wave Systems, I’m struck by the fact that this ambitious futuristic vision is coming out of one small Canadian company. Not only are they pushing quantum computing to unheard of levels of performance, but they’re also laying the groundwork for true sentient AI with learning algorithms that are already making mobile devices smarter and more capable. Quantum theory may deal in uncertainty as its stock and trade, but in this case, I think the odds of success are firmly on D-Wave’s side.