Google Buys Artificial Intelligence Start-Up DeepMind
Google has acquired London-based start-up DeepMind to expand further into the field of artificial intelligence.DeepMind describes itself as a “cutting-edge artificial intelligence company.” It combines techniques from machine learning and systems neuroscience to build powerful general-purpose learning algorithms.
The company was founded by Demis Hassabis, Shane Legg and Mustafa Suleyman. Its first commercial applications are in simulations, e-commerce and games.
DeepMind investors include Founders Fund, the venture capital firm run by former PayPal executive Peter Thiel, and Horizons Ventures, headed by Li Ka-shing, one of the richest people in the world.
Google reportedly paid more than $400 million for DeepMind, which would make it the company’s largest European acquisition so far. The company also reportedly beat Facebook to the deal.
Artificial intelligence, or AI, is a branch of computer science that aims to make computers behave more like humans, with capabilities such as reasoning, learning and planning. Google already uses this type of technology for many projects, such as its expanding language-translation services.
Google spokesman Tim Drinan confirmed the acquisition, but declined to comment further Monday. An e-mail sent to DeepMind seeking comment was not returned.
Google co-founders Larry Page and Sergey Brin have been focused on AI for at least a decade. Back in 2004, as the company was going public, the executives were already thinking about how Google search could be improved through this technology.
At that time, Page said Google search would be “included in people’s brains,” according to Steven Levy’s 2011 book about the company, In the Plex.
“When you think about something and don’t really know much about it, you will automatically get information,” Page explained.
Brin saw Google as a way to “augment your brain … you can have computers that pay attention to what’s going on around them and suggest useful information.”
“Eventually, you’ll have the implant, where if you think about a fact, it will just tell you that answer,” Page added in the book.
DeepMind published a study in December that claimed to show the firm’s AI technology learning to play some Atari computer games better than expert human gamers.
DeepMind’s technology performed better than humans on Breakout, Enduro and Pong, and it achieved “close to human performance” on Beamrider, according to the research paper.
The technology was far behind human performance when playing Q*bert, Seaquest and Space Invaders, “because they require the network to find a strategy that extends over long time scales,” the paper said.