Robots Moving Closer to Humans

As computers evolve, relationships between people and machines get complex.
Roboticists and animators have a term called the "uncanny valley" to describe that space where a human replica appears close to lifelike, but is just obviously fake enough to be creepy. R2D2 is fine because he's not pretending to be human; but the CGI "people" in Avatar or The Polar Express cause us to recoil.

Dr. David Hanson, CEO of Hanson Robotics, is trying to build the perfect robot. He says that "robots are increasingly autonomous in the way they interact with people socially." Since we're spending more and more time staring at square screens and other shiny, angular bits of technology, "it benefits people to humanize our technology," he said Tuesday at The Atlantic's Big Science Summit in San Jose, California.

Hanson's goal is to create robots with "character identity" that will form "true relationships with people." With a background in film and animation and three years at Disney under his belt, Hanson is creating shockingly lifelike robots that look like people and can even mimic human facial expressions.

And Hanson's dreams don't end with the face. He actually wants to create machines that "have human-like capabilities," maybe even human-level intelligence. To that end, Hanson and his team are developing open-source software in the hopes that the power of a million human minds can combine to create "genius-level machines."

But according to Wendy Ju, robotics researcher at the Stanford Human-Computer interaction group, robots don't have to be lifelike to be effective. In fact, you're probably using robots already and you don't even know it -- like automatic-opening doors. Ju has used unsuspecting pedestrians on the Stanford campus to experiment with doors of different intelligence -- doors that open the moment you get to them and close immediately after, doors that open fast and close slowly, doors that help you go through them without breaking stride.

Ju also found that people were twice as likely to use automated information kiosks if the kiosks actually waved at them using a mechanical arm as they walked by. Interestingly, she found that a waving arrow had just the same effect, so maybe it's the effect of physicality, not just the "human-ness" of a robot that's important.

Isaac Asimov once wrote a story in which a man was suspected of being an especially lifelike robot. One character dismisses the possibility, saying that humans are so well attuned to the characteristics and tics of other humans that "it would be impossible to palm something merely nearly right off on us." In the future, that may or may not be true. But as of now, the uncanny valley has yet to be bridged.

0 yorum: