uncanny

valley

Learn about what makes robot creepy

The Uncanny Valley is a term used in human-robot interaction to explain why robots sometimes appear to be creepy or unsettling. It was first proposed by Masahiro Mori in 1970 who explained the phenomenon with a graph. The graph is “human likeness” along the x-axis and “familiarity” along the y-axis; familiarity means how comfortable a human is and is a scale from “find it creepy” to “totally comfortable with it.”

 

Mori had observed that animated toys, puppets, robots, computer animation, all follow the same pattern of eliciting emotional reactions. If they are  fairly primitive in terms of how much like a human that they look, then they are not too creepy. Humanoid robots like Pepper look mechanical and thus signal that they are robots, so mentally they don’t creep us out. But more realistic human-looking robots like Ishiguro’s Geminoid robots (where he builds duplicates of his wife and daughter) are very creepy because while they have high human likeness in appearance, they don’t have corresponding high human likeness in actions. David Hanson’s robot Sophia often is displayed with electronics in the back of her head visible; this reduces the creepiness of a very accurate human face and body but slow and stilted interactions. Mori observed that the more human-like, the more comfortable (graph going up and to the right) up to the point where the appearance was very human-like but the actions or cognition didn’t match or there was one aspect that was just off (like eye movements) and that imperfection would create an overwhelming creepy response in a human. That’s the dip, or valley, in the curve. (Mori’s term - The Uncanny Valley- was probably mistranslated and should be The Valley of Eeriness see @liveScience)

The absolute best, and hysterically funny, explanation of the Uncanny Valley is in the Succession episode of the TV series 30 Rock where the writers try to explain to Tracy Morgan why his idea for an animated video game won’t work. Another fun exposition is http://tvtropes.org/pmwiki/pmwiki.php/Main/UncannyValley, which explains why Cybermen are scarier than Daleks. But the best implications of the Uncanny Valley is Brian Aldiss’ Super-Toys Last All Summer Long

September 17, 2018

Recommendation: Pick up a copy of The Caves of Steel and enjoy a true classic.

In 1953, Isaac Asimov spun an odd couple story about a human and robot detective team in  an over-populated future where humans on earth lived like naked mole rats, packed into cities. The ro...

July 9, 2018

Robots: Humanoids, both androids and more mechanical-looking service robots

What it gets right about robotics: the Uncanny Valley and schemas. Unlike HBO's Westworld, it does make the mistake of assuming emotions or something that can be added later.

Recommended watching...

April 28, 2018

HBO’s Westworld is located to the right of the uncanny valley, near the archipelago of governments proposing to regulate sexbots and strong artificial intelligence.

Here is a link to the article... http://robotics.sciencemag.org

-

-

Here is some of the reaction in...

April 23, 2018

HBO's Westworld is back with its cadre of robots who have been so well engineered to fake intelligence that they have it. Bad news for the endangered Delios Company Board of Directors but great news for us since the “fake it until you make it” approach is a reason to e...

March 5, 2018

Robots: Borg-like rectangular space ship robots, UAVs that look like Harrier jump jets, and Iron Giant style humanoid ground robots.

What it gets right about robotics: nothing, but there are some teachable moments about social robots and the Uncanny Valley.

Recommended w...

February 21, 2016

Robots: Androids, robot horses and rattlesnakes.

What it gets right about robotics: perception and the use of affordances, the manpower it takes to keep complex robots in operation, computer viruses

Recommended watching: Stop what you are doing and watch this movie now!...

Please reload

Index:

For further reading: