Turing Test

Alan Turing came up with the test for intelligence in 1950.
It comes form a game where a man sits in one room, a woman in a second, and a judge in a third.
The judge communicates with the people via teletype (text or chat now).
The judge had to determine who was the woman and who was the man.
The Turing test replaces one of the people with a machine, and the judge has to decide who is the person and who is the computer.
There are lots of recent partial solutions, but nothing really comes close.
The Loebner prize is an annual competition for this. There's a big prize for the system that first passes.
There is an annual domain specific test.
The recent claims of passing the test are really just weak methodology.
In the version we AI people generally consider a test for intelligence, the conversation can take as long as you like, and can be run multiple times.
The conversation can be about anything.