A Turing test is a test created by Alan Turing. The test is designed to test computational abilities. The idea behind the test is that there's a human trying to convince another human that they're human and a computer trying to convince that same human that they are also human. Can the computer win by getting the judge human to declare it the human rather than the human-human? If so, it passes the Turing test.
It's such a simple test...but what does it mean? If a computer can pretend to be a human well enough to fool a human, does that mean it's intelligent? And as intelligent as what? (Probably not a human, right? Humans can do things that are a lot fancier than pretend to be a human for a few minutes. But maybe it means that computer is smarter than a computer that couldn't pass a Turing test? I'm not sure of that either, though. There are a lot of cool things computers can do that don't involve pretending to be human. I think if I was a super sophisticated AI, I could find better things to do, for instance!)
Much of the difficulty depends on how long the computer has to pretend to be human. It wouldn't be hard to pretend to be a human for a few minutes, right? Even a very basic program be made to say something like this:
Computer: Hi! What's your name?
Tester: I'm Bill.
Computer: Hi Bill! It's great to meet you! What do you do for fun?
Tester: I like playing video games.
Computer: What a coincidence! I also love playing video games. What's your favorite game?
(Actually, I think that last conversation is more involved than the last one I had on a dating app! I kid, I kid...weirdly I don't remember the last time I tried out online dating, although it couldn't have been that long ago...I digress...)
So if the computer doesn't have to pretend for very long, probably any old program could do it. But the problem gets harder if the computer has to 'be' a human for hours. And what if instead of just typing on a black screen, the computer has to recognize voice and respond with auditory nuance? Or recognize a video feed? Or respond to that video feed with nuanced animations of its own? All of that's a lot harder than a simple chatbot...
Which is to say that this Turing stuff is really complicated! No wonder it's an ongoing (and fun!) debate! But it's cool to think that someday someone may crack the problem.
Comments