Google cloud chief sceptical of ‘AI singularity’
Thu 17 Nov 2016
Speaking at the Code Enterprise conference, Diane Greene, head of cloud business for Google, said that she does not believe that AI singularity will occur in her lifetime.
The ‘singularity’ refers to the point in time that machine intelligence overtakes human intelligence, a moment that many believe is rapidly approaching, in part due to recent progress in the fields of artificial intelligence and machine learning.
In his book The Singularity Is Near, author Ray Kurzweil predicted that technological singularity would take place in 2045. Kurzweil is currently the Director of Engineering at Google.
Greene, while acknowledging that artificial intelligence is progressing rapidly, remains unconvinced that the world is on the verge of the singularity.
“Nobody expected some of the advances we are seeing as quickly as we’re seeing them,” she said, “but, the singularity? I don’t see it in my sentient lifetime.”
She went on to note that “there are many things that machines cannot do that humans can do very, very well.”
A concern that goes along with fears of the singularity is the fear that as machine learning progresses, artificially intelligent machines will take jobs from humans.
Greene acknowledged that there was a risk that as AI becomes more useful, traditional employment categories could be affected. But she noted that as long as people were trained appropriately for the future, jobs would be available.
“I think it’s really incumbent on us to get the education out there to make sure everyone is digitally literate, because that’s where the divide is, and because if you’re digitally literate, you’re going to have jobs,” she said.
The Code Enterprise conference came just after Google announced the hire of two leading AI researchers to head up machine learning at Google. Fei-Fei Li, the director of AI at Stanford University, and Jia Li, head of research at Snapchat, have both accepted positions at Google.
Acknowledging that Google had hired many of the top researchers in the fields of AI and machine learning, Ms. Greene countered a question about whether Google had an inappropriate level of control over the future of AI by noting that many of Google’s products are open source and available to anyone who is interested in working with AI products.
TensorFlow, Google’s open source library for computation and data projects was released just over a year ago. And the company just set up a website called A.I. Experiments, where anyone from hobbyist to professional level can explore experiments created using Google CloudVision API and TensorFlow and use the tools to create their own applications.