Artificial intelligence (A.I.) and machine learning have the potential to help people explore space, make our lives easier and cure deadly diseases.
But we need to be thinking about policies to prevent the technology from one day killing us all.
That's the general consensus from a panel discussion in Washington D.C. today sponsored by the Information Technology and Innovation Foundation.
"When will we reach general purpose intelligence?" said Stuart Russell, a professor of electrical engineering and computer sciences at U.C. Berkeley. "We're all working on pieces of it.... If we succeed, we'll drive the human race off the cliff, but we kind of hope we'll run out of gas before we get to the cliff. That doesn't seem like a very good plan.... Maybe we need to steer in a different direction."
Russell was one of the five speakers on the panel today that took on questions about A.I. and fears that the technology could one day become smarter than humans and run amok.
Just within the last year, high-tech entrepreneur Elon Musk and the world's most renowned physicist Stephen Hawking have both publicly warned about the rise of smart machines.
Hawking, who wrote A Brief History of Time, said in May that robots with artificial intelligence could outpace humans within the next 100 years. Late last year, he was even more blunt: "The development of full artificial intelligence could spell the end of the human race."
Musk, CEO of SpaceX as well as CEO of electric car maker Tesla Motors, also got a lot of attention last October when he said A.I. threatens humans. "With artificial intelligence, we are summoning the demon," Musk said during an MIT symposium at which he also called A.I. humanity's biggest existential threat. "In all those stories with the guy with the pentagram and the holy water, ...he's sure he can control the demon. It doesn't work out."
With movies like The Terminator and the TV series Battlestar Galactica, many people think of super intelligent, super powerful and human-hating robots when they think about A.I. Many researchers, though, point out that A.I. and machine learning are already used for Google Maps, Apple's Siri and Google's self-driving cars.
As for fully autonomous robots, that could be 50 years in the future -- and self-aware robots could be twice as far out, though it's impossible at this point to predict how technology will evolve.
"Our current A.I. systems are very limited in scope," said Manuela Veloso, a professor of computer science at Carnegie Mellon University, speaking on today's panel. "If we have robots that play soccer very well by 2050, they will only know how to play soccer. They won't know how to scramble eggs or speak languages or even walk down the corridor and turn left or right."
Sign up for MIS Asia eNewsletters.