If watching the Terminator left you cowering behind the sofa, you are not alone.
When it comes to cyborgs, robots, and AI powered digital helpers the creepiest ones are those which look most like us.
A study about artificial intelligence has found ‘humanoid’ robots like CP30 in Star Wars are really quite likeable.
But if they look too much like people, robots tend to be disliked and mistrusted – perhaps because of a fear they could replace us in a dystopian future.
Our fears of certain robots even showed up in MRI brain scans, with activity spotted in ancient brain regions that deal with threats and social bonding.
If watching the Terminator left you cowering behind the sofa, you are not alone. When it comes to cyborgs, robots, an AI powered digital helpers the creepiest ones are those which look most like us (file photo)
They found people liked the human-like robots, with their strange, flawless faces, much less than robots which were clearly mechanical.
Dr Fabian Grabenhorst, senior author of the study, said: ‘People in our study were less likely to like or be willing to receive a gift from the most human-like artificial agents.
‘They can be seen as creepy because they just don’t look quite right.
‘These artificial humans trigger the part of the brain which responds to threats – an old part of the brain in evolutionary terms.
‘According to some theories, this may be because humans fear of being replaced by them.
‘Designers could consider this as they create robots. Robots are already being tested as companions in care homes in some countries, so it is important to understand how we respond to them as social partners.’
The study used pictures of real people, ‘mechanoid’ robots with only a primitive head shape, ‘humanoids’ like CP30 and androids like Data from Star Trek, as well as more modern artificial humans.
The ‘artificial humans’ looked flawless and smooth but with slightly exaggerated features such as larger lips or abnormally high cheekbones.
Rating each of the pictures for likeability, the 21 people in the study preferred robots which looked like humans – until they looked too human.
The artificial humans were the most unpopular, liked about half as much as pictures of real people in the study.
‘For a neuroscientist, the “Uncanny Valley” is an interesting phenomenon,’ added Dr Grabenhorst, from the department of physiology, development and neuroscience at the University of Cambridge.
‘It implies a neural mechanism that first judges how close a given sensory input, such as the image of a robot, lies to the boundary of what we perceive as a human or non-human agent.
‘This information would then be used by a separate valuation system to determine the agent’s likeability.’
The study about artificial intelligence found ‘humanoid’ robots like CP30 in Star Wars (left) are really quite likeable. But if they look too much like people, robots tend to be disliked and mistrusted – perhaps because of the fear they could replace us in a dystopian future
To see how much people aged 18 to 35 trusted the robots, they were asked if they would accept a present from them.
The presents ranged from a cinema voucher or a bottle of wine to dishwasher tablets and toilet cleaning products.
The results show about 90 per cent of study participants were willing to accept a gift picked by another person, but that fell to only around 10 per cent if it was ‘chosen’ by a human-looking robot.
People who rejected a present from an artificial human showed more activity in the amygdala – the part of the brain which weighs up threats.
Those who disliked artificial humans showed less activity in the medial prefrontal cortex, which is important in making friends.
The results suggest we do not want our artificial intelligence to look like the Terminator, or the unnerving ‘synths’ in Channel 4 show Humans.
Instead we are happier with old-fashioned androids, whose stiff posture and visible mechanical parts make it very clear they are not like us.
First author Professor Astrid Rosenthal-von der Pütten, from RWTH Aachen University in Germany, added: ‘Resembling the human shape or behaviour can be both an advantage and a drawback.
‘The likeability of an artificial agent increases the more human-like it becomes, but only up to a point – sometimes people seem not to like it when the robot or computer graphic becomes too human-like.’
The full findings of the study were published in the the Journal of Neuroscience.