By Daniel J. McLaughlin
Like Geppetto wishing that his wooden creation Pinocchio would be a real boy, humans creating robots wish the same thing. They are made to look almost like a human being, and they are hoping that they could act like a human being. There will be no wishing upon a star, however, as the technology evolves, and the dreams - and to some, a nightmare - come true.
Even in their early development, they reflect their creators both in physicality and process. Many robots are humanoid in shape with two legs and two arms. Computer scientists also use deep neural networks, to replicate the human brain and nervous system, in artificial intelligence; this is, however, flawed as the AI develops "catastrophic forgetting" (forgetfulness in skills and tasks they performed in the past when presented with a new task) - scientists at Google's DeepMind are looking to overcome this by basing their systems on neuroscience-based theories instead of the physical systems.
This catastrophic forgetting, however, may be charming to humans, and decrease the fear towards robots. "I'm only human" is a common excuse for people making mistakes, and if robots follow this behaviour and replicate humanity's tendency to mess up, the machines could endear themselves to humans. A recent study found that error-prone robots are more likable than error-free robots. Researchers monitored how participants responded to a robot that made unplanned errors, compared to a machine that flawlessly completed its tasks. Judging their behaviour and body language, they "responded more positively to the bumbling robot", Live Science reports, with the participants liking it "significantly more" than the robot that made no mistakes at all.
Writing for the Morgridge Institute for Research, Katherine Matthews and Bilge Mutlu argue that people tend to work better in environments familiar to them, and that's why they develop technologies that look or behave in recognisable ways. However, replicating human behaviour is a "very challenging and complex process", when researchers try to break things down into components such as speech, eye contact or movement. The problem they face is they do not replicate exact human behaviour, and instead create human-like behaviour. This "uncanny valley" is very close to human-like but it is not quite right; and it can come across as creepy. By being almost-human, rather than completely non-human, robots may be less trustworthy.
While the technology aims to imitate humans, it's currently not doing a particularly good job at it. For those concerned about being replaced by robots, they have not reached the stage of bettering humans. The Next Web calls robots that act like humans a "waste of time", arguing that they suck at being us. They should be made useful, rather than trying to imitate humans.
"One ridiculous robot after another graces our periphery in a never-ending cascade of failed human-interface designs. These almost-pointless machines are nothing but hyperbole factories and media magnets that somehow manged to keep getting created, despite engaging in borderline offensive behaviour," they argue.
Humans are both imaginative and lack originality. Whilst it is an act of genius to create robots and develop artificial intelligence, the design itself is likely to look familiar. They are designed by humans, and are humans in design. And, as the technology evolves, there is a hope that will be, one day, like a real boy.