When you see a robot, from the simplest roomba to the more charismatic and advanced Kismet artificial intelligence and expression system you still know deep down that a robot is a series of circuits and electrons. The facsimile of human behavior is intended largely at this stage in robotics development to comfort those with apprehensions about having robots around in the workplace or at home. But research at the University of Washington suggests that babies actually perceive robots as sentient already.
Perhaps one day we will enter a world where robots are either sentient or carry on a facsimile of sentience applicable to every day interactions. But in the mean time we’re stuck with the somewhat unconvincing simulations of human and animal behavior from our mechanical minions. But why do children not see robots as adults do now? Is it a youthful prescience of the world we may one day find ourselves in? Or is it a lack of understanding the difference between something that looks real and the genuine article? Or is it both?
The research suggests that babies can recognize the facial features of robots, even if they do not necessarily look anything like a human and have not been exposed to images showing the similarities between humans and robots. Babies can essentially pinpoint the features on the robot’s head and discover which of these features are similar to human eyes. Then, as it looks around the room they will follow its gaze. On the contrary, they will not follow the direction that a swiveling office chair is facing. Gaze following in this experiment was used as an indicator of human opinion of sentience.
It’s interesting to think that living in a world where sentient beings are seen on television all the time that do not actually exist, how this will affect our response to technological entities in the future. While it’s hardly likely that we are being purposefully “groomed” to accept robots into our lives by the entertainment we watch, it is a sort of self reflecting prophecy. As technology improves, so will it be able to show us ever more convincing robots. As a result we will become familiar with these robots. And it is really up to the viewer and those making the shows to decide whether or not humans will accept robots into their lives or keep them on the shelf and in highly regulated environments.
But with the way children react to a simple robot today, researchers have demonstrated how quickly a child’s consciousness can understand simple ‘symbolic’ representations of eyes on a flat metal surface to believe it is sentient. And if physical characteristics can represent eyes to a child so readily, is it also possible that easily programmed behaviors can do the same for adults? Perhaps AI experts shouldn’t be looking to create a truly conscious being, but a convincing enough replica for everyday use.
And how will such a being be received? While some would say it’s impossible to determine, a clue may lie in what Masahiro Mori calls the uncanny valley. The uncanny valley is the theory that as something approaches human-like appearance it goes through a period where “almost human” is disturbing (such as a corpse, a puppet, or a zombie.) And as it becomes a more perfect replica it once again is less disturbing. Perhaps behavior will be the same way.