The PS400m Megamind; Exclusive interviewA Huge Deal with Google Has Just Made Demis Hassabis One of the Country's Most Successful Entrepreneurs. the London Artificial Intelligence Genius Speaks for the First Time about How Clever Computers Can Cure Cancer and Why It's So Hard to Satisfy a Brain as Big as His

Article excerpt

Byline: Jasmine Gardner tech ed

WHICH one piece of technology would Europe's most valuable artificial intelligence expert be likely to have with him at all times, I wonder? Demis Hassabis looks at me with a hint of embarrassment. "I tend to be the sort of person who waits to see how useful a technology really is before I adopt it," he says, pre-empting the forthcoming reaction. Out of his pocket the man whose company Google has just paid PS400million to acquire pulls an old, slightly battered iPhone 5. It's not even the 5S. But what it is, is Apple.

"We'll have to sort you out," calls a voice from Team Google. That phone is a false idol in this house of worship. Someone at Google's Central St Giles HQ no doubt scrambled into action at that very second to find Hassabis a Nexus 5 (Google's flagship phone) pronto.

Hassabis is the 37-year-old founder and CEO of Bloomsbury-based DeepMind Technologies -- the threeyear-old neuroscienceinspired artificial intelligence (AI) company that hit the headlines this week when Google bought it.

In one swift move Hassabis has become one of London and the UK's most successful 'The thing I bought a computer. felt it this device' entrepreneurs and yet, until now, little was known about him or his mysterious firm. "We just like quietly getting on with our work," he says. He has barely been photographed since his teens. This is his first and only interview following the sale.

"It has been so crazy the last couple of days. It's a bit strange," Hassabis says, settling into one of Google's Seventiesstyle orange and white chairs. He is a neuroscientist, unaccustomed to this kind of attention. But this man, dressed head-to-toe in casual grey and black but for the flash of Day-Glo green on the topside of the arms of his spectacles, might just be able to change the world.

It is "within the last year," he says that "Google and Larry [Page, Google cofounder] found out about us. I got an email out of the blue to come and have a meeting."

DeepMind has been building "learning algorithms -- ones that automatically learn how to do things from raw data, rather than being programmed to do things". Its work is not easy for the layman to get to grips with. "We are not trying to copy or interface with the brain," explains Hassabis. "We look at state-of-the-art neuroscience and cherrypick the key principles behind how we think the mind works and see if we can convert them into an algorithm."

The only public demonstration so far of what his company has achieved was getting his AI to learn how to play Space Invaders on an old Atari computer just by showing it the information on screen. "It turns out games are a perfect testing ground for trying these things out in a controlled setting."

Sounds clever, but this is just a start.

More sophisticated AI will rest on understanding more of the brain's mechanisms. "Some people believe that the brain is based on one master algorithm, but it's much messier than that and an artificial system will need multiple algorithms to have all those capabilities."

Hassabis says that he and his team are "quite hard to impress, we see stuff like Siri [Apple's "intelligent" personal assistant] and we think, 'We could do so much better than that'," but as yet his company is lots of science and "core technology that could be deployed in interesting ways" but not a lot of finished product. Which makes the purchase by Google all the more remarkable. It seems to be a price for, literally, Hassabis's head.

was I was magical And why not? His must be one of the greatest brains in the country, perhaps the world. He was a champion chess-player in his early teens, finished his A-levels at 16, built a multi-millionselling videogame by 17, got a double First in computer science from Cambridge, a PhD in cognitive neuroscience at UCL, published research in neuroscience that was listed as one of the top 10 scientific breakthroughs of 2007 and became an academic at MIT and Harvard. …


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.