Google continued a spree of recent acquisitions earlier this week with its £400-million acquisition of the London-based artificial intelligence company, DeepMind, founded by chess pro and games designer Demis Hassabis.
There’s no doubt about it: Google is expanding its view of software’s role in the world, venturing into self-driving cars, humanoid robots and health care. But its DeepMind acquisition is, in many ways, just more business as usual.
“Google buys cutting-edge technology when they think there’s good tech and good people,” said long-time Google watcher Ezra Gotthiel, a senior analyst at Technology Business Research. “I think they have that sense of ‘I’m just adding to my portfolio, I just want to get better and better products and people, and I have all this money sitting around that otherwise I just have to invest at low interest rates.’”
DeepMind’s version of artificial intelligence is particularly well suited to search. (Indeed Re/Code, formerly All Things D, reports that Google will fold DeepMind staff into its search team.) And to fund “moon shots” like its forays into robotics, the company absolutely has to stay ahead of its competitors in search, Gotthiel told Singularity Hub.
“The things they could do with AI always are good understandable heuristics that look like intelligence, but that really underlie what they try to do all the time. Search and Google Now are the kind of things that AI technology gets us closer to,” Gottheil said.
DeepMind also filed a patent for a method that could improve Google’s image search results by allowing the user to use visual descriptions rather than keywords. And according to its website, the company was also building e-commerce applications for artificial intelligence — potentially like Amazon’s plan to guess at what users will buy in order to deliver it faster. (Fingers crossed we won’t soon be lugging boxes to UPS to return items we’ve only thought about buying.)
Google could additionally be looking to remedy its previous failure in useful commercial search — signaled by its move to turn Google Shopping into a search engine exclusively devoted to paid advertising, Gotthiel noted.
Of course, with IBM targeting health care as a major market for its AI tool, Watson, Google could also be eyeing smarter searches that would allow medical researchers — potentially those employed at its own health spinoff, Calico — to make better sense of huge, clumsy databases.
For a tech aficionado, most of these use cases may sound prosaic, but companies have to go where the money is first. And iterative improvements often reach a tipping point at which they represent a radical change. If Google uses DeepMind’s AI to improve search, for example, in the short run the company may unveil little more than a nifty image search tool and a beefed up Knowledge Graph database. But in the long-term, the same work could deliver a computer that can answers any question, returning spoken answers rather than URLs. Factor in that computers are becoming small enough to challenge our notions of freestanding machines and are increasingly being deployed in humanoid robots like those made by Google’s other recent purchase, Boston Dynamics.
With that in mind, it’s clear that Google’s business as usual and the most futuristic visions of what the company could bring to market are not so far apart after all.