Sunday, February 2, 2014

Google’s Grand Plan to Make Your Brain Irrelevant



Google is on a shopping spree, buying startup after startup to push its business into the future. But these companies don’t run web services or sell ads or build smartphone software or dabble in other things that Google is best known for. The web’s most powerful company is filling its shopping cart with artificial intelligence algorithms, robots, and smart gadgets for the home. It’s on a mission to build an enormous digital brain that operates as much like the human mind as possible — and, in many ways, even better.
Yesterday, Google confirmed that it has purchased a stealthy artificial intelligence startup called DeepMind. According to reports, the company paid somewhere in the mid-hundreds of millions of dollars for the British outfit. Though Google didn’t discuss the price tag, that enormous figure is in line with the rest of its recent activity.
Lifelike robots, sentient machines, the Jetsons’ smart home in the sky. Google is spending billions to make itself the place where these fantasies become facts.
The DeepMind acquisition closely follows Google’s $3.2 billion purchase of smart thermostat and smoke alarm maker Nest, a slew of cutting-edge robotics companies, and another AI startup known as DNNresearch.
Google is looking to spread smart computer hardware into so many parts of our everyday lives — from our homes and our cars to our bodies — but perhaps more importantly, it’s developing a new type of artificial intelligence that can help operate these devices, as well as its many existing web and smartphone services.
Though Google is out in front of this AI arms race, others are moving in the same direction. Facebook, IBM, and Microsoft are doubling down on artificial intelligence too, and are snapping up fresh AI talent. According to The Information, Mark Zuckerberg and company were also trying to acquire DeepMind.

The New AI

Google’s web search engine already uses a powerful type of artificial intelligence to find what you’re looking for in the chaos of the web, and it has built an insanely profitable ad business atop this engine. But recently, the company has been bulking up its roster of geniuses as it seeks to explore a new branch of artificial intelligence known as “deep learning.”
Basically, the idea is to mimic the biological structure of the human brain with software so that it can build machines that learn “organically” — that is, without human involvement.
Google is already working to apply these insights to its familiar consumer products and services. Deep learning can help recognize what’s in your photos without asking you to tag them yourself, and it can help understand human speech, a key tool for its smartphone apps and Google Glass computerized eyewear. But Google also sees the new AI as a better way to target ads — the core of its business.
The DeepMind acquisition is one more step down this road. And though the company has not said as much, you can bet that this new form of AI will also play into things like Nest smart thermostats, the Google self-driving cars, and its big push into robotics.

A Century of Sci-Fi Dreams Come True

At the moment, it seems, no other institution on earth has the concentration of brain power — coupled with the money, technology, and freedom — to chase the dreams that have fueled a century of science-fiction speculation. Lifelike robots, sentient machines, the Jetsons’ smart home in the sky. Google is spending billions to make itself the place where these fantasies become facts.
In a profile of deep-learning pioneer and now part-time Googler Geoff Hinton, WIRED’s Daniela Hernandez writes that the key difference between deep learning and other approaches to artificial intelligence is that it aims to free machines from the need for human intervention, to give them a human-like understanding of our environment. By building so-called neural networks that approximate the brain, Hinton and company are trying to make it possible for Google to understand language, speech, and the physical world without having to be told what its machines are seeing, hearing, or touching.
For many of us, Google already functions as an important part of what WIRED columnist Clive Thompson has called our outboard brain. The more Google “knows,” the less we have to remember. We just Google it. Now imagine that same kind of intelligence Google applies to the web set loose on your personal existence, not just online but out in the real world.
If its artificial intelligence dreams come true, Google might end up knowing you better than you know yourself. As we export more and more of our intelligence to Google, the question might become: What are our own brains for?
Marcus Wohlsen is a staff writer for Wired Business and the author of Biopunk: DIY Scientists Hack the Software of Life

No comments:

Post a Comment