When we talk about the definition of the Uncanny Valley, the hypothesis that explores the degree of discomfort humans experience when viewing replicas or anthropomorphic representations, they must always be accompanied by photographs or similar images. what are we going to see,
Because although AI is undoubtedly advancing by leaps and bounds, introducing itself into all kinds of areas of our lives, the ability to reproduce perfect images of us humans has not yet been achieved. have been found. In fact, the pictures below are one of the most recurring problems AIs face when they try to “copy” us.
The footage in question surfaced a few days ago after midjourney enthusiasts shared AI-generated footage of parties, none of which actually took place., Let’s have a look at some such pictures:
There is no doubt about the progress and achievement of producing these images out of nothing but something that produces a certain disapproval upon viewing them, the description or failures Which turns these “people” into things more like “monsters” than “humans”. That skin color, those hands whose fingers are impossible or the monstrous number of teeth.
Yes, at first glance the pictures may look like a normal party, but everything falls apart as soon as we start counting fingers or teeth. In general, people were not impressed with the effect, the reason? According to Explained A designer and AI expert:
Due to the complex geometry of hands, there is no standard set of lines or shapes that AI can recognize as a hand. To generate a realistic hand, the AI needs to combine many different shapes and arrangements. There are about 30 geometric deviations in the human hand from the length and width of the fingers to the metacarpals and wrist joints.
Put another way, AI works on pattern recognition and pattern generation, rather than having the concept of a hand or a tooth.
The Uncanny Valley Continues To be present [IFLScience]