You know those movies where the machines take over and the human race gets enslaved by computers? Pure applesauce, says Eric Schmidt, chairman of Google’s parent company, Alphabet (GOOG).
We have little to fear from the rapid growth of artificial intelligence, he said at an event at Columbia University in New York on Monday. Fears about computers run amok are the stuff of movies, he argued, and that the technology serves to help people not hurt them — including when it comes to things like income and employment.
“I worry about inequality but there’s no evidence the stuff we do creates a permanent underclass,” said Schmidt, in the course of an interview with Columbia’s dean of journalism, Steve Coll.
He added that critics have fretted about technology’s impact on the job market for decades but that, overall, tech has created “millions and millions” of new jobs.
The timing of Schmidt’s comments was perhaps ironic since, only a week ago, a Google computer drew on its machine learning prowess to defeat the world’s top human player in an epic match of the game Go. The computer’s 4 games to 1 victory was another milestone for machines, coming five years after IBM’s Watson super-computer routed two Jeopardy champions on national TV.
Schmidt noted that AI software (which Google and others are making open source) is serving to automate tasks like sorting photos or even driving a car, but argued that only very low skill, repetitive type work will be affected. He acknowledged this includes some types of journalism jobs, including certain sports reports and corporate earnings stories, but said this trend will be limited.
“Creative jobs and the caring jobs [like health care] are the ones that are robust against everything,” he said. “There’s no evidence that the world I live in is displacing that.”
Beyond the impact of machine learning on the job market, Schmidt also downplayed its military and cyber-war implications. In his view, as the world’s computers become more inter-connected, AI may come to serve as a “defensive shield” for the global network by identifying and isolating abnormal activities.
“This technology may be very pro-defense. We don’t know yet,” he said, but acknowledged that Google’s leaders did worry about AI in the hands of dictators and authoritarian regimes.
At a time when hacking is regular front-page news, including a recent cyber-plot against a dam in New York, many in the room did not appear to be reassured by Schmidt’s observations. One questioner asked how to avoid the all-controlling computers of the movie Minority Report.
In the end, though, Schmidt’s most convincing answer might have been the simplest one: It’s too difficult.
According to Schmidt, AI has made enormous progress in recent years, but has now reached a huge computational roadblock. The upshot is that AI remains very good at solving problems—but only after humans have done the hard work of defining the problem in the first place. This was the case with Google’s Go computer, whose game-playing feat took teams of people two years to program.
For more about Google, watch:
As such, our fears about machines taking over may still be as far away as they were at the time of the 1968 movie, 2001: A Space Odyssey. The villain in the film, by the way, was a computer named HAL whose name was rumored to be based on a big tech company—not Google but another firm with the nearby three-letter acronym of IBM. (IBM)