Remember those Charles Schwab commercials from the last year or two ago? They used an animation technique called rotoscoping, where artists trace each individual frame of a live action scene. I recall finding those commercials annoying, bordering on downright disturbing.
It turns out that this reaction is pretty common, so much so that the phenomenon was given a name that has stuck for the last 40 years: the “uncanny valley,” a phrase coined by roboticist Masahiro Mori. I recall first reading about this in an article from the December 2008 issue of Scientific American. More recently, I saw it just last week in The Straight Dope. (If you haven’t read the Straight Dope, I recommend it; Cecil Adams’ columns not only have some interesting information, but they are usually a pretty fun and amusing read.)
The uncanny valley is simple to describe. Consider plotting a graph of humans’ emotional connection with a robot versus the extent of similarity of said robot to a human form. (Wikipedia has a detailed such graph here.) Such a graph will be increasing, with the exception of a marked “valley” or dip just prior to the point of indistinguishability between a robot and a human. Put more simply, as robots, or animated images, or whatever, become more and more life-like, we appreciate that similarity more and more… but when a robot or image is so life-like that it looks “almost but not quite human,” we find it off-putting or even disturbing.
Beyond the Chuck Schwab commercials mentioned earlier, I think movie-makers might also find the uncanny valley an important consideration. I remember Final Fantasy: The Spirits Within being an interesting movie at least from a technological perspective… but it also seemed a little spooky. The more recent Beowulf had a similar effect, and not just because Angelina Jolie had a tail. The interesting question, I think, is why? What causes this reaction in us? And if it’s not actually human, how close to “human” must a robot be to climb out of the uncanny valley?