Nadella: AI Needs Direction
In the great debate over whether the robots will save us or destroy us, Microsoft CEO Satya Nadella is staking out a more activist position.
Nadella revisited the artificial intelligence issue in a speech at The Economic Club of New York on Wednesday. "I feel like sometimes we in tech, even, abdicate control: '[AI] is going to happen tomorrow and our best case is that we're going to be domesticated cats or whatever,'" Nadella said.
His comment references pessimistic statements like those from scientist Stephen Hawking, who told the BBC, "The development of full artificial intelligence could spell the end of the human race...It would take off on its own, and redesign itself at an ever-increasing rate." Or SpaceX/Tesla founder Elon Musk, who has written, "The risk of something seriously dangerous happening is in the five-year timeframe. Ten years at most."
Nadella countered Wednesday that "no, it's a choice. I'm not making fun of that as a consequence. It could happen, but only if we abdicate."
He acknowledged the reality of downsides, such as unintended consequences of automation, especially job displacement. In Nadella's view, however, the eventual behavior of AI depends on the values and actions of the people in the tech industry. "We as a society -- starting with Microsoft -- have to do some of our very best work at skilling...students in school or people who are displaced midcareer."
Those comments piggyback on a major theme of his recent book, "Hit Refresh," in which Nadella dedicated an entire chapter to the future of humans and machines.
As he wrote there, "We can't seem to get beyond this utopia/dystopia dichotomy. I would argue that the most productive debate we can have about AI isn't one that pits good vs. evil, but rather one that examines the values instilled in the people and institutions creating this technology."
At Microsoft, Nadella wrote that he is pushing the company's substantial AI-focused workforce to follow principles that AI must be "designed to assist humanity...be transparent...maximize efficiencies without destroying the dignity of people...be designed for intelligent privacy [and] have algorithmic accountability."
Posted by Scott Bekker on 02/08/2018 at 1:34 PM