Tim Hornyak

Jun 18th 2018

How Humanlike Will AI Robots Like Sophia Become?


In his latest novel “Origin,” Dan Brown hints at how artificial intelligence — and AI robots — could become avatars for real people. Symbologist Robert Langdon finds himself embroiled in an adventure pitting science against religion, but he has the help of a sophisticated AI named Winston. This software represents slain futurist Edmond Kirsch and comes across as a combination of Apple’s Siri and Jarvis from the “Iron Man” franchise; it’s also able to adapt human traits with remarkable accuracy.

In that respect, this piece of science fiction isn’t too far ahead of science fact. The past 10 years have seen a renaissance in AI research and development as the power of computer processors and sensors has increased while cost has gone down. Smart, useful virtual assistants like Siri are now in the pocket of everyone who owns a smartphone.

Rise of the Androids

Sophisticated, humanlike robots are now serving as platforms for the latest AI tech, which couples natural-language processing skills with machine learning and the vast information resources available online. Some of the most lifelike have been developed in East Asia. Japan has produced a range of startling androids, such as the Geminoid series developed by Osaka University researcher Hiroshi Ishiguro. Hong Kong-based Hanson Robotics, however, has stirred controversy since the 2016 debut of Sophia, a female android that has been shown off at numerous high-profile meetings and interviews, drawing global interest.

With her exposed cranial circuitry and knack for eyebrow-raising statements, Sophia is slightly unnerving in appearance and performance. She has cameras in her eyes, the ability to mimic human facial expressions and carry on a conversation. She was recently outfitted with a humanoid robot body, giving her rudimentary walking skills. Company founder and CEO David Hanson said in an interview with The Telegraph that locomotion can give machines the ability to “understand what it means to feel human … to feel with us. That’s really important for artificial intelligence, I think, to be safe in the future.”

Whether or not robots can ever “feel” is one question, but Hanson envisages a time when robots could be indistinguishable from humans. While many robotics observers doubt that will ever happen — even questioning the degree to which Sophia’s answers are scripted — the robot was awarded Saudi Arabian citizenship last year, said Forbes; she was also named an “Innovation Champion” by the United Nations Development Programme.

Do Asimov’s Three Laws Still Apply?

Should AI robots have the right to citizenship? This calls into question whether they — or their creators — can be responsible for their actions. Indeed, the hype surrounding Sophia recalls novelist Isaac Asimov’s Three Laws of Robotics, which state that robots must not harm humans, they must follow human orders and they must protect their own existence. Seventy-six years after they were introduced in Asimov’s 1942 short story “Runaround,” the Three Laws of Robotics are increasingly relevant today: the recent death of an Arizona woman after a collision with a self-driving Uber car reignited debate about how to protect people from autonomous machines that are becoming more and more common in public places, said The New York Times.

It’s clear that smart machines will require a far more complex regulatory framework than Asimov’s laws; in 2014, law professor John Villasenor suggested that product liability laws present the best hope for effective legislation. Meanwhile, despite the stagecraft of Sophia the robot, we’re still a long way away from a time when robots might have the general intelligence and physical skills that one would expect of a robot butler, though companies like Japan’s SoftBank Robotics and its Pepper robot are attempting to get there.

Ultimately, it’s more worthwhile to consider Sophia on her own terms — an intelligent machine that may reveal a thing or two about how our own minds work, along with how we act toward our fellow humans.

“Her brain and mind are not extremely close emulations of the human mind, and we don’t necessarily want it to be,” Hanson Robotics’ chief scientist Ben Goertzel said in an interview. “As she gets more and more intelligent, I think she can be more rational and also more wise and compassionate than humans are.”