Can artificially intelligent robots be our friends? Our helpmates? Our companions? Roboticists and AI researchers are trying to make it so – and the first fruits of their labors are about to come onto the market. But there are already hints that the efforts will touch some of humanity’s hot buttons.
Take Hanson Robotics, for example: Its latest creation, Sophia, combines an AI chatbot with an expressive humanlike face. She can talk enthusiastically about helping humans in health care, education and customer service. But she can also go off script.
“Do you want to destroy humans? Please say no,” roboticist David Hanson, the company’s founder, asked Sophia during a CNBC interview at this month’s South By Southwest technology conference in Texas.
“OK, I will destroy humans,” it replied. “No, I take it back!” Hanson said with a laugh.
Closer to home, a Microsoft teen chatbot named Tay was hijacked by mischievous Twitter users and transformed into a foul-mouthed racist, less than 24 hours after it was released onto the Internet. Microsoft had to take Tay offline, delete the offending Tweets and try resetting its AI attitude.
Such problems shouldn’t be surprising to science fiction fans, who have been pummeled by robo-dystopias ranging from the classic 1927 film “Metropolis” to last year’s “Ex Machina.” But despite the challenges, scores of companies around the world are working on robots that are meant to have the smarts and the actuators necessary to interact with humans in everyday environments.
Why? Tandy Trower, the founder of Seattle’s Hoaloha Robotics, says robots will help humanity address a coming demographic crisis: the graying of the global population.
“This isn’t about geek appeal about robots,” Trower told GeekWire in an email. “It is an opportunity to address an increasingly serious challenge of how we can fill the gap between an increasing senior population and a supply of resources that will not be able to keep pace.”
Census figures suggest that America’s over-65 population will nearly double by the year 2050, to 83.7 million. In other countries, such as Japan and German, the graying trend is already well under way. It’s no accident that those are countries where robot caregivers are on the rise. Robots are expected to help meet the needs of all those seniors in situations where human caregivers aren’t available or affordable.
Senior citizens aren’t the only humans facing a potential caregiver gap, said Oren Etzioni, CEO of Seattle’s Allen Institute for Artificial Intelligence. He said the rise of rehabilitation robots hints at the shape of things to come.
“Most of us can’t afford to have high-quality personalized care,” he told GeekWire. “If we could all afford to have a personal trainer, that would be a wonderful world. But the health care system won’t support that.”
A Merrill Lynch report estimates that the global market for personal robots, including “care-bots,” could rise to $17.4 billion by 2020.
Robots that look human
Some of the robots being readied to meet the demand look eerily human: In addition to Hanson’s Sophia, there’s Nadine, the robo-receptionist modeled after its creator, Swiss-Canadian computer scientist Nadia Thalmann; the conversational Geminoid robots created by Japan’s Hiroshi Ishiguro; and the human-mimicking Actroid robots, also under development in Japan.
Hanson told GeekWire that he’s gearing up production in Hong Kong for a line of expressive facial robots, plus the operating system and software development kit to go with them, for introduction to the market later this year. The robo-heads can be placed on mechanical bodies for applications ranging from Disney-style animatronic exhibits to personal companions.
How far will these robots go?
“I believe that robots will become people,” Hanson said. “I believe that in time they will develop the complete capability of a human, to understand us, to have general intelligence and the willful desire to grow and reach their potential the way that humans experience it. I don’t know when that’s going to happen. That’s our goal. It might be 10 years, it might be 15 years before we reach that goal.”
Does that include love and sex? Hanson said sexbots are likely to happen, but his company is “not moving in that direction.”
“I believe that debases the potential for robots,” he said. “It’s a kind of trap.”
Little robot buddies
Trower takes a different tack. He and his colleagues at Hoaloha Robotics have been working on sociable robots in stealth mode for more than five years – at first, strictly on the software, but now on the hardware as well. Trower expects to begin user testing by the end of the year, but he’s in no hurry to get a product to market. (He told GeekWire that he’s not yet ready to reveal the robot’s design.)
In Trower’s view, it’s not a good idea to make an intelligent robot too humanlike. He subscribes to the theory of the “uncanny valley” – that is, we’re so fine-tuned to recognizing other humans that interactions with an almost-but-not-quite human entity would feel super-creepy. For that reason, Hoaloha’s child-sized robots won’t look like us, although they’ll have a display that can register social cues.
“We also do not present our robot as your all-knowing oracle,” Trower said. “Instead, it presents itself as an eager assistant that gets smarter with interaction with the user. So rather than the Jetsons’ Rosie, we are more like the scout character, Russell, in “Up”. Our goal is that this better sets user expectations and ensures that the robot is never perceived as the user’s superior.”
Other robots of this ilk include SoftBank’s Pepper, Blue Frog Robotics’ Buddy, Fraunhofer’s Care-O-Bot and MIT robotics pioneer Cynthia Breazeal’s desktop Jibo (which looks a cross between Amazon Echo and R2D2).
Trower said Hoaloha’s offering will be mobile and interactive. It may not fetch your car keys, but it will help you remember where the car keys are. In fact, Trower sees memory assistance as one of the key capabilities that an older population will be looking for in their care-bots.
“One might argue that I’m on a mission to create a piece of technology that I see myself needing someday soon,” the 63-year-old said, half jokingly. “I have to admit that I’m hoping they get done with the driverless cars before they take my driver’s license away.”
Humans and robots together
Will there ever come a time when artificially intelligent robots bridge the uncanny valley and blend in with the human population? “Probably not in my lifetime,” Trower said.
But Hanson sees that as a necessity.
“We shouldn’t fear crossing this gulf. We should explore it,” he told GeekWire. “We need to humanize our machines. We need to teach them to care about us. The biggest risk comes from super-intelligent machines that don’t understand, or care.”
Luminaries like SpaceX’s Elon Musk and British physicist Stephen Hawking may worry that intelligent machines will eventually turn on us, but the Allen Institute’s Etzioni doesn’t see that as a huge concern.
“We’re much more likely to have robot slaves or robot assistants than robot overlords,” he said. “The overlord thing is overblown.”
Ready or not, the intelligent machines are already among us. They’re beating us in the game of Go. They’re diagnosing our diseases. Soon they’ll be driving our cars. But what form will our most intimate robots take? That’s up to us humans.
GeekWire’s Alan Boyle will be the moderator for an MIT Innovation Forum panel on the future of artificial intelligence and robotics from 5:30 to 8:30 p.m. PT March 30, at Impact Hub Seattle, 220 Second Ave. S. Panelists include David Hanson of Hanson Robotics (dialing in remotely from Hong Kong); Mark Hammond, founder and CEO of Bonsai AI in Berkeley, Calif.; Mark Greaves, technical director for analytics at the Pacific Northwest National Laboratory; and Illia Polosukhin, an AI researcher with Google Research. Admission is $39, or $29 for students.