[While some are excited by the possibilities for human-like robots, others point to possible negative consequences of this form of ‘medium as social actor’ presence; this story from Tech Insider includes the full 58:06 minute panel during which Stuart Russell made his comments (which begin at the 21:35). For a cogent review of the status of AI see Slate’s coverage, which includes a link to a recent interview with Russell. –Matthew]
[Image: Meka Robotics. Source: YouTube]
Human-like robots may have a disturbing impact on actual humans
Danielle Muoio
May 4, 2016
Researchers hailing from NASA to Google have invested in humanoid robots.
There’s the Dreamer robot that was acquired by Google-turned-Alphabet in 2013. And there’s NASA’s humanoid robot designed to help astronauts in space.
But one robotics expert is warning against the creation of humanoid robots — specifically, those that closely resemble humans.
Stuart Russell, vice chair of the World Economic Forum Council on robotics and artificial intelligence, called for a “ban of highly human-like humanoid robots” during the Milken Institute’s panel titled “Artificial Intelligence: Friend or Foe?”
“We’re just not equipped in our basic brain apparatus to see something that’s perfectly humanoid and not treat it as a human being,” he said. “So in some sense, a humanoid robot is lying to us using the lower levels of our brain we don’t get to control.”
Russell’s viewpoint echoes a controversial theory known as the Uncanny Valley that postulates the more a robot begins to resemble a human, the more uncomfortable or revolted people become.
A 2015 study published in Cognition gave some credence to the theory — participants were asked to rate 80 pictures of social robots as well as play an economic investment game to measure their comfort and trust levels for each robot. The researchers found that the robots’ likability sharply decreased once they started looking extremely human-like.
“30 Rock” actually explains this really well [in a 0:55 minute video on YouTube].
Russell said he thinks the uncanny valley could pose a major issue for children in particular.
“Particularly for young children, growing up in a household where there are humanoid robots and humans it could be extremely confusing,” he said. “And we could see psychoses developing as a result of machines not behaving as the child expects them to behave because they think it’s a human.”
A study done by various Japanese researchers actually found that children are likely to show “serious abusive behaviors” towards robots. The researchers concluded that the more human-like the robots looked (or more they approached the uncanny valley) the more likely it was for kids to start beating them up.
You can watch Russell make his comments [on YouTube].