Pop culture has generally presented the image of a fully functioning robot as one that is on par with humans. It has two legs, two arms, two hands, two feet, a torso and a face with two eyes and a mouth. Perhaps it’s because it is human nature to like robots more when they look more familiar and human-like. Perhaps it just sells movies. What I want to know is how far away are we from this integrated community of humans and robots living side by side? Can we build robots that walk and talk as we do? Maybe better than we do?
From my research of the web, two terms seem to be popping up quite frequently, and they are the ‘Turing Test’ and ‘Uncanny Valley’. Now while both of these sound relatively harmless, their definitions are delightfully terrifying.
I’ll start off describing what the ‘Turing Test’ entails. Basically it’s a test to see if a computer is smart enough to fool a human into believing that the computer is also a human. What’s most scary about this test is that computers have actually passed it. Yes, sometimes ignorance is bliss!
The test is over 65 years old, but in 2014 the test was finally passed for first time by the computer program ‘Eugene Goostman’, which emulates a 13 year boy.
The test is quite simple. A human interrogator is isolated. He/she must then ask a series of questions to both a computer and a human. After numerous tests, the interrogator must decide based on the responses which they believe to be a computer and which to be a human.
To pass the test, the computer needs only to be mistaken as a human 30% of the time. It’s fair to say that those chat bots that keep bothering you on Facebook probably didn’t pass.
‘Uncanny valley’ is different to ‘the turing test’ because it focuses more on the physical similarity between humans and objects. When we focus in on the robots that fall under this term we get some interesting results.
Take Geminoid androids for instance:
This particular robot was made using a mould of his creator and is programmed to emulate his body language and gestures. The creator has even implanted his own hair into the robot.
This android is just one of many androids created by Hiroshi Ishiguro Laboratories, each of which are equally creepy.
Another intriguing robot that I stumbled upon on the internet was Jules.
Jules is a lifelike robot created by Hanson Robotics with extremely advanced animatronics throughout his face and high res cameras in his eyes that allows him to ‘lock onto his conversational partner’.
Jules is so fascinating because the cameras allow him to record interactions and learn from them. So if you’re nice to Jules, expect a warm reception in return. If you’re not nice to Jules, expect him to act appropriately.
So to summarise, this robot is able to pass as a human, track people, learn from its mistakes and it can hold grudges. Sound a little familiar? That’s because I just described the exact characteristics of the Terminator.
Another robot designed for human interaction is Actoid, who has well designed AI that allows him/her to converse freely with humans in 4 different languages. They were designed to answer any questions posed in over 2000 different ways and are programmed with over 40,000 phrases.
And like Eugene, Geminoid, Jules, and Actoid there are dozens of different computers and robots out there that are being developed more and more human-esque every day. So the idea of being best friends with a robot like Bender from Futurama within our lifetimes isn’t unimaginable.
In the second part to this blog, I hope to look at the non-physical side and explore how Artifical Intelligence has begun to emulate human thinking in ways quite unimaginable.