Why has Google bought an AI company?
- 28 January 2014
- From the section Technology
Earlier this week Google spent £400m buying a UK firm that specialised in artificial intelligence.
DeepMind is a privately-held company founded by Demis Hassabis, a 37-year-old neuroscientist and former teenage chess prodigy, along with Shane Legg and Mustafa Suleyman.
According to DeepMind's website it builds "powerful general-purpose learning algorithms".
But what does Google plan to do with its newly purchased expertise?
The company has not issued a detailed statement on the acquisition, but that has not stopped industry observers from trying to work out the motives behind it.
Google has recently bought a number of robotic companies and artificial intelligence is often associated with the development of these machines.
Several news websites asked if Google was planning to combine its robot and AI purchases and build "Skynet" - a reference to the fictional, self-aware artificial intelligence system that features in the Terminator films.
But the tech news site Re/code reported that the DeepMind team would report to Google's search team and not its robot division.
According to the site's sources, the team at the AI company were working on "a game with very advanced game AI, a smarter recommendation system for online commerce, and something to do with images" before the acquisition.
DeepMind had trained software to play video games without teaching it the rules. It had published a paper explaining how computers could learn to beat expert players of several Atari games.
Previous projects that Jeff Dean - who heads the team at Google that DeepMind will join - has worked on might give clues to how Google would use its new talent, reported Re/code.
In one project, a network of 16,000 computers taught itself to recognise photos of cats after analysing 10 million stills from YouTube videos. Another involved a "deep network" used for the "purposes of large visual object recognition tasks and speedy speech recognition".
Alternatively the founders of DeepMind could find themselves working along side Ray Kurzweil who joined Google in 2013 reported TechCrunch.
He is one of the "most prominent individuals associated with the singularity movement" which is the theory that "human beings and artificially intelligent machines will sync up to push innovation forward at an unprecedentedly fast rate."
He has said that he wanted to create a search engine so sophisticated that it could answer questions without you actually asking.
It appears that if DeepMind's expertise is used for search then the company wants assurances in place that it will not be abused. The Information news site reported that Google had agreed to set up an ethics board to make sure this did not happen.