Google Facebook

Google Beat Facebook for DeepMind, Creates Ethics Board

Google, which is acquiring DeepMind Technologies, has agreed to establish an ethics board to ensure the artificial intelligence technology isn't abused, according to two people familiar with the deal.

The unusual step of establishing the ethics committee comes as Google, which is paying more than $500 million to acquire the company, won the deal over Facebook, several people familiar with negotiations said. Facebook was in serious acquisition talks with DeepMind late last year, these people said, and it is unclear why talks fell apart.

DeepMind's technology aims to make computers think like humans and has been used in demonstrations of computers playing video games. Like many other innovations, the technology could be used to controversial ends. The DeepMind-Google ethics board, which DeepMind pushed for, will devise rules for how Google can and can't use the technology. The structure of the board is unclear.

A spokesman for Google confirmed the acquisition and declined to comment further. A spokesman for Facebook could not immediately be reached for comment. The Information reported in December that both companies were looking at DeepMind in their race to bulk up on experts in the hot field. Re/Code earlier reported that Google was acquiring DeepMind for $400 million.

DeepMind, which is based in London, has amassed one the biggest concentrations of "deep learning" experts in the world. Deep learning is part of a branch of artificial intelligence called machine learning. Deep learning tries to teach computers to see and hear the way humans do. The secretive company, which is backed by Founders Fund and Horizon Ventures, has about 30 PhD’s in computer science, according to one of the people involved in the deal.

The deal come as Google and Facebook have been battling over talent in multiple areas of technology, including search, recommendations and personalization. These areas are important for building new generations of services that think and act more like people do, including virtual assistants like Google Now and robots. Machine learning contributes by helping computers recognize speech and physical objects, including human faces, and by providing recommendations to individuals based on information about them.

DeepMind was working on machine learning algorithms that would be used by specific commercial industries, but it hadn't announced a major consumer product. At a December artificial intelligence conference, the company showed off an algorithm that could learn how to play video games, such as the Atari game “Breakout,” without previous knowledge of the game. Several conference attendees, including Yoshua Bengio, an artificial intelligence professor at the University of Montreal, said they were impressed the demonstration.

As of December, DeepMind had about 75 employees, including more than a dozen deep learning graduates, according to a person who worked there. 

Amir Efrati is executive editor at The Information, which he helped to launch in 2013. Previously he spent nine years as a reporter at the Wall Street Journal, reporting on white-collar crime and later about technology. He can be reached at [email protected] and is on Twitter @amir