One of the main questions in AI, is "Are we there yet?". How can we test if computers are intelligent, and what does that mean anyway?

In the 1950's Alan Turing devised such a test, called a Turing test. Its goal is to determine if a computer is intelligent. Basically a judge talks in a chat session to a computer and a human, and has to determine which is which. If they cannot, then the computer has passed the test (or perhaps the human has failed...).

At first glance it seems like a pretty stupid test, but if you consider a judge who is intelligent and understands AI, and a human who is also intelligent and answers the questions to the best of their knowledge, then it is actually a very good test. Of course if the judge is not so intelligent, or does not understand AI, or is more interesting in chatting then figuring out which is the computer, then it is a pretty easy test to pass, using a simple ELIZA chatter bot. Also if the human is dumb, or crazy, or a child, or doesn't speak the same language as the judge, the test is also easy to pass.