People have for many years worried about computers and robots becoming self-aware and somehow overthrowing us but it seems perhaps we don’t need to be quite so concerned; our weapon in ensuring that the global hypercomputer grows up to be nice? Cats.
Ok, bit of an exaggeration. We’ll need more than cats.
But anyway, Google have again created something impressive in its research labs; a virtual neural network that turned out to be like a simplified version of the human visual cortex and when it was presented with a data set consisting of images from YouTube it decided, with no training or input from programmers, to learn how to identify and define a cat and was able to generate an image of a cat based on general features. The result is influenced by our love of filming our feline friends and sharing their antics with the world but is still a great achievement in the field.
In the future computers based on virtual neural network type structures like this would be able to intelligently search and process the vast stores of information we have and will continue to pour onto the internet. Look forward to the day when you ask your computer for directions to Bedford and it replies “later, can’t you see I’m having a really bad day”.
It should be noted here and now that I reserve the right to use a cat picture to illustrate any article on this blog where there is even a vague relevance.