Le 26/02/2021. If you work remotely – either because of the COVID-19 pandemic or because that is the nature of your...
© Ned Snowman / Shutterstock.com
© Albanpix / LtdREX
In previous blog entries, we’ve talked about the general reaction of the public when they released trailers for the live-action film ALADDIN and musical film adaptation of Andrew Lloyd Webber’s CATS. In general, people watched both trailers and found them both to be a little…creepy. This phenomenon has a name: the Uncanny Valley.
A Japanese robotics professor, Masahiro Mori, wrote about the concept in an essay whose title translates to ‘valley of eeriness’. He wanted to explore why humans seem to have a universal, uneasy feeling when looking at wax figures (androids had not yet been invented). The term ‘Uncanny Valley’ was first mentioned in a book from 1978. But what is it, exactly?
The basic premise is that when we see automated depictions of humans, we like them when they are not at all realistic. We also like them when they are hyper-realistic. When Mori’s put his research in the form of a graph, he found that likeability/familiarity drops at a specific point on the graph. The horizontal axis measures realism and the vertical axis measures the ‘likeability’ or ‘familiarity’ of the depiction. We see the curve rise steadily as we move gradually from realistic through the spectrum until a specific point. At that point, likeability drops dramatically, but rises again the closer we get to hyper-realism. This creates a visual ‘valley’ of the graph’s progression.
Scientists are trying to figure out why this happens. The most common idea is that the robots that are ‘just a little bit off’ can remind us of humans – specifically psychopaths. Simply put, we are more disturbed by something that is a little bit wrong than we are by depictions that are not at all humanoid. Scientists noted that facial expressions of the robot are also key. For example, if a robot is smiling with its mouth but has angry eyes, it triggers alarm bells in humans. We can tell that something is wrong – maybe not consciously – but we have the capacity to sense it. The example of smiling with angry eyes appears to a human as incongruous, as if the robot is trying to “suppress deep-lying emotions.” This has led people to wonder if creating truly human-like robots is a good idea, asking, “What if we make them too good, and we can’t tell if they are human or not?”
The good news is that scientists are starting to think that humans will always be a little bit ahead in the ‘robot vs human’ perception, so we will always be capable of distinguishing human from robot… or will we?
uneasy = gêné
trailer = une bande-annonce
previous = précédent
to be released = (ici) sortir
both = tous les deux
creepy = qui donne la chair de poule
uncanny = troublant
eeriness = (ici) une ambiance inquiétante
wax = la cire
depiction = une représentation
likeability = (ici) quelque chose d'attachant
to drop = (ici) chuter
to rise = monter
closer = plus près
to figure out = (ici) comprendre
simply put = pour faire simple
key = (ici) la clé
to trigger = déclencher
to sense = ressentir
to lead = mener [qqn] à faire
truly = vraiment, réellement
to suppress = réprimer
deep lying = (ici) profond
to tell = (ici) distinguer
ahead = en avance