When Stephen Hawking predicts humanity’s downfall, it’s no joke. He is, after all, an expert on quantum mechanics and general relativity. So in May, when the world-renowned theoretical physicist co-authored an editorial in the UK’s The Independent warning that continued advances in artificial intelligence could be “our worst mistake in history,” people paid attention. They read, eyes glued to the screen, as Hawking suggested that blindly building thinking machines was analogous to inviting alien civilizations to Earth without considering that they might just want to incinerate us. And then those readers politely rolled their eyes.
“I think Hawking is conflating autonomy and intelligence,” says Oren Etzioni, a former University of Washington computer science professor, founder of multiple startups (including airfare predicting software Farecast), and current CEO of the Allen Institute for Artificial Intelligence. (Of course the guy who’s building smart computers wants to downplay their potential for bringing about the apocalypse.) AI2, as the Paul Allen-backed company calls itself, is busy teaching machines how to literally read text and then take tests on the material. Etzioni envisions a day in which “intelligent computers” can augment human learning. “Our goal,” he says, “is to have them collaborate with us to solve major problems for the human race, from climate change to cancer.”
Despite what Stephen Hawking thinks, when the robots rise up, they'll be more likely to hold our hands than crush us under their steel heels.
That’s right: Despite what James Cameron—or Stephen freaking Hawking— would have us believe, when the robots finally rise up they’ll be more likely to hold our hands than crush us under their steel heels. And one of the biggest birthing suites for advanced robot tech is right here in Seattle. Beyond AI2, Amazon is fulfilling customers’ orders at warehouses around the country with an army of robots that could grow to 10,000 by the end of 2014. And UW’s computer science department has spun off a handful of projects, including BluHaptics, which is developing telerobotics (basically really advanced remote controlled bots) that can aid in deep-sea exploration. “Between the university, the big companies, and the startups, there’s a significant robotics and artificial intelligence industry and academic research community here,” says Ryan Calo, a robotics law professor at UW. And the one thing stopping that community from being bigger isn’t (human) brainpower. It’s funding.
“With some investment from Microsoft, we could be a serious contender.” Fittingly, Seattle’s most promising startup was founded by a Microsoft expat. Robotics researcher Tandy Trower left Redmond in 2009, after his biggest advocate, Bill Gates, retired. The next year he launched Hoaloha Robotics, and he’s been hard at work on a mobile personal-assistant robot for elder care ever since. “Think of it as a PC on wheels,” Trower says. More specifically, a vaguely humanoid PC that can navigate your house, seek you out, and verbally remind you to take your medicine.
But it goes even further than that. “As we age, we become more socially isolated,” he says. “And there’s tons of research that says the degree to which you have an active social network determines how healthy you are.” So Trower and his team are building in a series of communication algorithms that will allow the robot to inform patients about their family members’ social media moves, read email, commiserate about the weather, and even tell jokes. Believe it or not, it’ll also ask for feedback and ditch sub-par joke material.
Through those interactions, it will gather information about the patient’s well-being (sleep patterns, appetite, activity level) that can be passed on to authorized medical professionals. Though, as Trower will admit, incorporating that kind of adaptive technology is the biggest challenge and the main reason he’s still two or three years away from bringing the product to market.
“Dr. Hawking’s fears about [the power of] artificial intelligence seem very distant for me,” he says, bemoaning the glacial pace of advances. “I kind of wish that weren’t so true, because it would make my job a lot easier.”
In February 2011, after famously losing a three-day Jeopardy! tournament to IBM’s Watson supercomputer, Seattle’s own Ken Jennings jokingly proclaimed, “I for one welcome our new robot overlords.” At the risk of sounding like a co-conspirator in Skynet’s inevitable takeover, if their worst offense is telling the occasional bad joke, we’re right there with you, Ken.