Seven Things Google Might Do With a Quantum Computer

The Citizen's Guide to the Future
May 16 2013 5:17 PM

Seven Things Google Might Do With a Quantum Computer

Quantum computing
In theory, quantum computing could solve some problems exponentially faster than traditional computing.

agsandrew / shutterstock.com

Google and NASA are the proud new owners of a quantum computer, which they will use to launch an artificial-intelligence lab at NASA’s Ames Research Center in California. Google announced the initiative, called the Quantum Artificial Intelligence Lab, in a blog post today: “Our goal: to study how quantum computing might advance machine learning.”

Will Oremus Will Oremus

Will Oremus is Slate's senior technology writer.

That sounds impressive. But what, exactly, does it mean?

Advertisement

As Greg Kuperberg explained in Future Tense last fall, a quantum computer is a new type of supercomputer that seeks to harness “quantum randomness” to perform operations far more efficiently than traditional computers. The one Google and NASA bought was made by a company called D-Wave, and there is some debate about whether it’s really accurate to call it a quantum computer. Nonetheless, it recently solved at least one optimization problem 3,600 times faster than a conventional machine, raising the possibility that it could indeed lead to advances in machine learning.

That’s a big deal for Google, because machine-learning algorithms form the underpinning of the company’s most exciting new technologies. Basically, they’re procedures that allow computers not just to perform a specific set of predefined tasks, but to “learn” from their mistakes and user feedback and get better at those tasks over time—or even acquire the ability to perform new ones.

These algorithms are stunningly powerful, but they also tend to require massive computing power. For example, last year Google ran a cutting-edge machine-learning experiment in which computers learned to identify cats on YouTube—without being told in advance what cats looked like. That experiment required 16,000 computer processors running in parallel. As you can imagine, 16,000 processors are not going to fit onto your Android phone, let alone your Google Glass.

If Google can tap the theoretical power of quantum computing, those types of operations could get much easier. The company did not say precisely how it might use them in the real world. But off the top of my head, here are just a few of the Google technologies that could conceivably benefit from improved machine-learning algorithms:

  1. Google Now, the mobile personal-assistant program that attempts to anticipate your needs
  2. Self-driving cars
  3. Google Goggles, which recognizes images like the Eiffel Tower when you point your phone’s camera at them
  4. Search by Image
  5. Voice Search
  6. Google Prediction API
  7. And, of course, good old Web search

That’s just for starters, though. For more ideas, check out Google’s vast caches of published research on machine learning and machine translation. Meanwhile, Google’s blog post alludes to quantum computing’s potential to tackle problems as far afield as protein folding and climate modeling. And who knows what NASA’s engineers might have in mind?

It’s possible that the D-Wave 2 will end up solving problems that Google and NASA haven’t even thought of yet. Or it might turn out to be a total bust. But that’s a luxury that both NASA and Google can afford, for different reasons. Here’s to moon shots.

Future Tense is a partnership of SlateNew America, and Arizona State University.