The Turing Test
- In 1950, Alan Turing wrote a paper on a test for intelligence.
- The test is based on a game.
- A judge sits in a room while two contestants (a computer and
a person) sit in
one or two other rooms.
- The judge can communicate with the contestants via a
teletype (but there are some interesting ideas about
communicating via video link).
- The judge holds a conversation with each of the contestants,
and his goal is to find out which is the human and which
the computer.
- If the judge can not tell or guesses at chance (50% correct)
then the computer is intelligent.
- We can argue whether this is fair or not, but it is clear
that no artificial system currently passes this test, and
- any system that could pass it, would be really good.
