An android update, for you techno-paranoiacs
That humankind will one day find itself battling robots that we ourselves created has been a staple of science fiction for years (not to mention a go-to gag for writers of The Daily Show, who fret even about the gizmos wielded by CNN anchors). One version of the scenario is that the robots will look like us. So how close are scientists to developing robots that can pass for humans? In the latest issue of Perspectives on Psychological Science, in an article titled "Human-Android Interaction in the Near and Distant Future," a psychologist and a professor of computer scientist at the University of Illinois at Urbana-Champaign provide a handy update.
"In reviewing the current state of various robotic technologies," write Neal J. Roese (the psychologist) and Eyal Amir (the computer scientist), "it becomes clear that several fundamental hurdles make it unlikely that the androids of 50 years from now will be indistinguishable from human beings."
Phew. But while that general finding will relieve some people, even as it disappoints others, the article's main interest lies in the details. Five decades hence, robots should have no problem negotiating city streets with aplomb, the authors suggest, either by walking or other means. The technology already exists, although engineers need to improve the durability and robustness of robo-locomotion.
Perhaps surprisingly, the robots of 2060 should also be able to recognize the emotions that flash across human faces, Roase and Amir predict. "This currently is an active and successful area of … research." Moreover, given the state of knowledge about human facial-muscle action, it could well be feasible to create androids that mimic human expressions.
At least from a distance: one hurdle in android tech remains the creation of realistic skin and hair. Even in 50 years, Roese and Amir say, android faces will be identifiable as such from several meters away.
Of course, the development of artificial intelligence remains the largest barrier between current knowledge and "the threshold of indistinguishability." And advances in AI have been "glacial," Roese and Amir observe. They foresee a world, in 2060, in which robots "may have a conversational capacity slightly more complex than the automatic telephone services used by corporations today." Not quite the "Blade Runner" scenario some have feared.
We won't reach that vaunted threshold of indistinguishability for at least a century, Roese and Amir conclude. In the meantime, they say, psychologists have much to contribute to the creation of androids. This includes such small matters as teaching them how long they can stare at humans, and how close they can stand next to them, before the humans start to freak out.
Yes, that's about as far as these scholars' dystopian imaginations go: in the future, hyperintelligent androids might invade our personal space. Never let it be said we weren't warned.
