More Turing. This one from 2007:
I've been thinking about writing something on definitions of artificial intelligence since Christmas when my niece received a "toy" called 20Q. If you have not seen it, it is stunning. Seriously, check it out, it IS that weird. About the size of a tennis ball, it is a computerized version of the game twenty questions. You think of something and after twenty "yes or no" questions, it takes a guess.
My reactions to the toy were probably pretty standard. It went through three phases: (1) Hey, pretty cool, (2) oh my god, (3) this is eerie. When it guessed moustache, I was impressed. When it guessed head of lettuce, I was knocked out. But when it guessed electrical outlet, I was worried that it was some sort of NSA bugging device. This thing is unbelievable.
It is the result of an experiment in neural nets which the designer put on the web, so that it could play for 24 hours a day. Every game, it learns, making new connections. It makes "neural connections" in much the same way the human brain does and establishes a web of beliefs. As a result -- and this is what really impressed me -- it will guess correctly even if fed some wrong information. A portion of the resulting network of connections is put on a chip and forms the "mind" of the toy.
My niece was describing the way she and her friends at school reacted to the thing. They started making disguised hand gestures so that the toy couldn't see them or hear them in case there was a microphone or camera inside of it. At that point, an image of Alan Turning popped into my head. The machine had passed his test.
In 1950, Alan Turing published a paper in Mind called "Computing Machinery and Intelligence" in which he sets out the first definition of artificial intelligence. It is a phenomenological definition, that is to say, a definition based on human experience. The idea is that if a person interacts with the machine and after a significant amount of time mistakes it for a human interaction, then the machine is intelligent. These kids knew it was a machine -- they put in the batteries -- but still thought it had to be connected to a human mind in order to do what it did.
I was reminded of this last week when Lindsay and Aspazia both linked to the gender genie. It is a program designed to take a bit of text and determine whether the author is a male or a female. Amusing and cool, I'll grant you, but the real joke is somewhat obscure.
You see, in the 1950 paper, when Turing sets out to describe the test, he motivates the discussion by sketching out what he calls the "imitation game," something that could be played at parties.
The new form of the problem can be described in terms of a game which we call the 'imitation game'. It is played with three people, a man (A), a woman (B), and an interrogator (C) who may be of either sex. The interrogator stays in a room apart from the other two. The object of the game is to determine which of the other two is the man and which is the woman.In other words, you take a man and a woman and put them in a room with a typewriter. One of the people, but you don't know whom, sits at the typewriter and types out responses to written questions slipped under the door. (The typewriter keeps you from being able to identify handwriting.) It might go something like this, according to Turing,
Q: Please write me a sonnet on the subject of the Forth Bridge.
A: Count me out on this one. I never could write poetry.
Q: Add 34957 to 70764
A: (Pause about 30 seconds and then give as answer) 105621.
Q: Do you play chess?
A: Yes.
Q: I have K at my K1, and no other pieces. You have only K at K6 and R at R1. It is your move. What do you play?
A: (After a pause of 15 seconds) R-R8 mate.