Tuesday 18 October 2016

How real can VR be?

In Arthur Conan Doyle's story "The Adventure of the Mazarin Stone" (published in 1921), Sherlock Holmes fools some criminals into revealing the whereabouts of a diamond by convincing them that he is playing his violin in the neighbouring room. In fact, he is hiding behind some curtains listening to their conversation while a gramophone record plays the Hoffman 'Barcarole'.

Some years earlier, in 1895, at the Grand Café in Paris, the Lumière brothers presented film of a train arriving at a station.  Legend has it that, as the train loomed large on the screen, the audience panicked and ran away screaming.  Almost certainly though, the accounts are as fictional as Holmes' trickery with the gramophone.

Both technologies record and reproduce aspects of reality.  What would it take though, for a virtual reality to be mistaken for a real reality?  Should there be a VR version of a Turing test?  If a listener were placed outside two booths, one containing a real violinist and the other playing a recording of the violinist, would the listener be fooled?  Could a projection be displayed beside a closed window in such a way that someone in the room could not tell which showed the outside world?

Even if a technology could pass such a test when new, could it continue to do so?  Cutting edge technology quickly becomes blunt. CGI special effects that, 20 years ago, seemed impressive, now seem clumsy.  Arguably, the same question could be asked of the Turing test for artificial intelligence (AI).  If the AI did not learn in the same way as humans, then it may not consistently pass the test.