Article by Michael Szollosy
“The desire to anthropomorphise, the need to connect, is powerful, and that is why this thing is going to sell.”
So says Daniel Graystone, inventor and CEO of Graystone industries in the American network series Caprica. The prequel to the 2004 remake of Battlestar Galactica, Caprica tells the story of how the genocidal Cylons came into existence. Graystone is trying to develop a robot for use by the military, but realises that his will be more successful if his robots look and act like human beings. First, it needs to be pointed out – evidently with some frequency – that bipedal robot soldiers are probably the most inefficient way that robots can be used in military combat, and not at all what a truly sophisticated artificial intelligence would use to take over the planet and enslave the human race.
But given that, Graystone makes a very important point: there is a very deeply-rooted impulse to anthropomorphise – to attribute human qualities to things that are not human – and this seems to be a big factor in the development of human-robot interactions.
This became apparent in a recent post on this blog about the race to create ‘personal robots’: Amidst some genuinely beneficial devices and some really exciting innovations, we found some very ambitious promises about just how much these robots will serve not just as useful machines but also as ‘companions’. Sometimes, these robots involved little more than putting an animated face on a baby monitor, or putting a tablet on wheels and adding a soft female voice. (Why these robots are so often anthropomorphised as female is perhaps something that must be addressed in another post.)
But these robots, looking and sounding increasingly human, look as though they are going to sell.
The impulse to anthropomorphise is an irrational drive, sometimes leading us to draw some strange conclusions. It has little to do with the instrumental utility of a device. A good hammer is effective at putting nails into wood. Would a hammer with a personality, with a face, be more effective at that task?
Well, maybe, yes, as it turns out. As we humans are instinctively social animals, perhaps there are some clear benefits to robots with whom we can interacts on a social level: robots with material bodies, instead of disembodied intelligences, and, perhaps best of all, robots with faces, with whom we can more naturally interact.
The impulse to anthropomorphise is an important part of human evolution, and plays an important part of our learning. As Tony Belpaeme explains, this impulse can be seized upon to massively improve the effectiveness of technology in applications like learning and in caring, especially with children.
However, the impulse to anthropomorphise also leads to some assumptions about robots that are unrealistic and, in some cases, dangerous. If one is presented with a robot that has a face, one automatically, instinctive, makes assumptions about those things of which the robot is capable. One might expect, for example, that the face staring back at us shares our intellectual capacity, or our ability to empathise. (This might be in some way what is responsible for the phenomenon known as the uncanny valley, where one experiences a degree of discomfort when in the presence of a life-like humanoid robot.)
Anthropomorphisation, more worryingly, might lead us to expect that robots share human abilities to exercise judgement, for example, in combat situations. In a 2013 TEDx lecture, Noel Sharkey describes how military planners, having seen impressive killing machines, make all sorts of promises about how robot soldiers will be able to autonomously identity and eliminate targets. But these planners have no conception of the serious perceptual and intellectual limitations of robots, let alone their complete lack of moral agency, emotional engagement or critical faculties in the exercise of judgment.
Noel Sharkey – Toy Soldiers to Killer Robots
Looking at a robot with a cute cartoon face, or even a mean-looking Schwarzenegger look-alike, one might assume – automatically, unconsciously – that robot capable of all sorts of human behaviours, feelings and thoughts of which it is simply not capable. And that’s even before the marketing men and overly keen programmers (with an overestimation of their abilities) make their promises and videos that seduce us even further. As ever, what is needed is an informed discussion, and some careful thinking how to effectively and intelligently use our tendency to anthropomorphise, not exploit it.
Tony Belpaeme – The power of robots with a face