Uncanny Valley: Psychology and How We See Robots

Standing just over four feet, Honda’s ASIMO has shaped something like an astronaut, the two cameras which serve as his eyes concealed behind a shiny black visor, and does not have trouble with the ‘uncanny valley’. ASIMO can walk upstairs, avoid obstacles, deliver trays of coffee, learn to recognize objects, faces, and vocal cues, and even conduct symphonies and connect to the internet to give up to the minute weather reports. The newest model represents the cutting edge in bipedal robot technology and has given rise to a number of Honda innovations in mobility, including a set of mechanized braces patterned after his legs that assist real people with disabilities.

He is also the humanoid robot that many subjects in studies of robot-human interaction seem to be most comfortable with. Unlike more realistic automatons, ASIMO is specifically designed to be evocative of human beings but not a perfect reproduction. If his technology is so advanced and painstaking, why not add a face? Why not skin?

Why are Robots Scary?

Masahiro Mori, a Japanese robotics researcher, went straight to the root of the problem, tackling what he termed the “uncanny valley” effect in the 1970s. He posits that human beings become increasingly more comfortable with robots as they appear more familiar; for example, ASIMO, whose white plastic coverings and black visor remind us of human NASA officers, is easier to interact with than a six-legged “spider” robot used in rescues, whose wires, joints, and actuators are showing.

The same is true for androids — robots specifically designed to mimic humans in as many aspects as possible — but to a point. Mori points out that as robots get closer and closer to perfect human emulation, people tend to notice minute differences that curb curiosity and cue the creep factor.

What is the Uncanny Valley?

The uncanny valley refers to a dramatic dip when levels of human observer comfort versus levels of android human-like behaviour and appearance are plotted on a line. The line quickly increases in steepness around fifty percent likeness- about where ASIMO lies in the spectrum of perceived likeness. Just after sixty percent or so, the line sharply drops off into negative familiarity values and soars back up again at closer to eighty or ninety percent likeness. That dip in the valley where most androids fall at the moment since the technology does not yet exist to simulate every single facet of natural human movement. At the very lowest point of the valley are the sorts of androids that people describe as frightening- those which are nearly perfect but whose eyes or faces seem “dead” or whose mechanisms are readily apparent.

Fear that Robots are Going to Take Over the World

Jules, an extremely realistic android built by Hanson Laboratories for the purposes of exploring the emotional and conversational capabilities of robots, doesn’t have the back half of his skull or hair, leaving the wires in his head exposed. Even though his eyes are lively and his face is nearly spot-on, this makes the interaction all the more strained, since according to Mori’s theory, observers draw a parallel between a lifelike face and their own and imagine the wires in their own heads.

Fears of human bodies being modified, abused, or reanimated in the name of technology are exploited again and again in Hollywood, along with the idea that the robots will one day, Terminator-style, render us obsolete.

The Uncanny Valley in Cyberspace

Mori’s effect is observable outside the lab- a search for a video of Osaka University’s android Repliee Q2’s unveiling on YouTube produces a range of comments along the lines of “they are gonna take over one day that’s when the world will end.” Conversely, many of ASIMO’s videos garner comments about how cute he is or how amazing his artificial intelligence has become. The majority are vastly positive.

One video of the little robot missing a step and taking a tumble is titled “Poor Asimo.” ASIMO is non-threatening and child-like, and there is no misapprehension about the fact that he is a robot. As such, people interacting with him treat him differently from his silicone-skinned cousins. Though Q2 and Jules both elicit a high level of emotional response in a robot from those who work with them, outsiders are less likely to be at ease with them.

“Resistance is Futile” – Robots, Artificial Intelligence, and Androids

Hollywood-style robots of nightmare are little at this point but fiction. All three robots are capable of learning on various levels, though ASIMO still can only recognize up to ten faces stored in his databanks, Q2 must be taught human movements using sensors, and Jules rather philosophically ponders his lack of emotional range.

What makes their intelligence threatening or amazing lies with what views we as observers impose on them. Until every jerky actuator movement is smoothed out or every potential companion is designed to be specifically comforting, humans will likely continue to see doomsday — and morbid curiosity — in every realistic electronic reflection of themselves.

realistic looking robot
Leave a Reply

Your email address will not be published. Required fields are marked *