Botworld
While we’re on this topic, let me show you another example of something that I’m sure is absurd but we’ve just got used to it.
This is a screenshot from 20Q, a 20 questions game online. You know, you think of an object, and the other person asks questions: is it animal, vegetable or mineral. Is it bigger than a microwave, can you find it on a farm, that kind of thing.
20Q’s artificial intelligence is a of a type called an “artificial neural network.” It’s a computer program that learns. Most computer programs are written much like a recipe. We know exactly what’s going on. A learning program, like 20Q, we don’t: it’s congratulated when it guesses correctly, and punished when it isn’t.
But let me read you a quote from the game’s designer:
“Every once in a while 20Q hits you with a question that seems completely off the wall. It doesn’t think the way a human thinks. As a human being, our strategy tends to be get a vague idea of what it is, focus in on one object and try to prove or disprove it. The 20Q [artificial intelligence], however, can consider every single object it knows simultaneously, so with each question you answer, certain objects become a little bit more likely to be what you’re thinking of, and certain objects become a little bit less likely. It then chooses a question that will cut the number of likely objects in half.”
[Source: interview with 20Q’s designer.]
Playing 20Q, it doesn’t seem to be getting anywhere. It asks you 20 random questions. And then, bang, it makes a guess, and it’s usually right.
The website of 20Q is doubly weird. Because it’s a learning program, every time you play it, it gets better at playing. And it’s been learning for 17 years, and millions and millions of games.