Don’t be evil, genius: Google buys a British artificial-intelligence startup

Don’t be evil, genius: Google buys a British artificial-intelligence startup

Feb 1st 2014 | From the print edition

WHEN a search engine guesses what you want before you finish typing it, or helpfully ignores your bad spelling, that is the result of machine learning, a branch of artificial intelligence. Although AI has been through cycles of hype and disappointment before, big technology companies have recently been scrambling to hire experts in the field, in the hope of building machines that can learn even more sophisticated tasks.

IBM said this month it would invest $1 billion in a new division to develop uses for Watson, its computer that understands human language. But this week Google cemented its lead in this field by paying around £400m ($660m) for DeepMind Technologies, a startup in London that has yet to announce a product.

The boss of DeepMind, Demis Hassabis, previously created video games such as “Evil Genius” and “Theme Park”, in which the events and characters can respond credibly in an enormous variety of potential situations because of an artificial-intelligence algorithm humming in the background.

DeepMind’s 75 geeks will join the world’s leading group of machine-learning experts, which Google has been assembling in the past few years. Google’s main source of income, its search engine and the accompanying ad-placement system, is driven by machine learning. The firm’s self-driving cars rely on it, as do the intelligent thermostats made by Nest, a firm it has just taken over, and the robots made by Boston Dynamics and other robotics outfits it has been buying.

The technology is already the backbone of many other internet firms. It is why Facebook and LinkedIn have that slightly creepy ability to find people you know, and why Amazon and Netflix are good at suggesting books and films you might like. It also helps intelligence agencies to identify terrorist networks.

As machine learning leaves the lab and goes into practice, it will threaten white-collar, knowledge-worker jobs just as machines, automation and assembly lines destroyed factory jobs in the 19th and 20th centuries. For example, the technique has been applied by researchers at Stanford University to tell whether a biopsy of breast cells is highly cancerous, something that until now has required a human expert to assess.

Another of DeepMind’s founders, Shane Legg, has predicted that artificial intelligence running amok will be the biggest existential risk to humans in this century. Its founders have asked Google to set up an “ethics board” to consider the appropriate use of machine learning in its products. The creator of “Evil Genius” is ensuring that his new overlord sticks to its motto, “Don’t be evil”.