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. 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."
Robert D. Atkinson, president of the Information Technology and Innovation Foundation, said people are being "overly optimistic" about how soon scientists will build autonomous, self-aware systems. "I think we'll have incredibly intelligent machines but not intentionality," he said. "We won't have that for a very long, long time, so let's worry about it for a very long, long time."
Even so, Russell said scientists should be focused on, and talking about, what they are building for the future. "The arguments are fairly persuasive that there's a threat to building machines that are more capable than us," he added. "If it's a threat to the human race, it's because we make it that way. Right now, there isn't enough work to making sure it's not a threat to the human race."
There's an answer for this, according to Veloso.
"The solution is to have people become better people and use technology for good," she said. "Texting is dangerous. People text while driving, which leads to accidents, but no one says, 'Let's remove texting from cell phones.' We can weigh this danger and make policy about texting and driving to keep the benefit of the technology available to the rest of the world."
It's also important to remember the potential benefits of A.I., she added.
Veloso pointed to the CoBot robots working on campus at Carnegie Mellon. The autonomous robots move around on wheels, guide visitors to where they need to go and ferry documents or snacks to people working there.
"I don't know if Elon Musk or Stephen Hawking know about these things, but I know these are significant advances," she said. "We are reaching a point where they are going to become a benefit to people. We'll have machines that will help people in their daily lives.... We need research on safety and coexistence. Machines shouldn't be outside the scope of humankind, but inside the scope of humankind. We'll have humans, dogs, cats and robots."
