Human or Computer? Take This Test. Discusses the ways you can have an automated process identify a computer vs. a human. Not a Turing test, although they are discussed, but rather a gatekeeper/checkpoint. A human has no problem identifying a word in distorted text or sound; but a computer - currently - has little or no chance of figuring it out. This does have some issues for disabled users and text-only browsers; but for the most part the benefits are worthwhile.
Another interesting point in this article is this... We can make a computer "smart" at maths because it's a relatively set body of knowledge. 2 + 2 = 4, big brother aside. But identifying a word in distorted text is something we don't understand nearly as well; so we can't express the process in code well enough for the computer to do it. Advances are being made; but to create a true artificial intelligence, we will have to understand ourselves far better than we do now.
Post a Comment