Quick! What’s the Turing Test?
C’mon, ask me a hard one, why dontcha? Everyone knows what the Turing Test is:
… a human judge engages in a natural language conversation with two other parties, one a human and the other a machine; if the judge cannot reliably tell which is which, then the machine is said to pass the test.
Well, according to AI guru David Fogel (CEO of Natural Selection, Inc., which stays in business applying genetic algorithms to real problems) everyone knows what ain’t so.
In fact, Turing proposed a different test. Someone is communicating with two entities, one of which is a woman. The other is either a man, or a construct. If the construct does as good a job of imitating a woman as the man does, then the construct is intelligent.
Fogel went on to trash the whole concept, arguing that bamboozlement capacity wasn’t a very plausible measure of intelligence; he prefers to focus on the capacity to make decisions in support of objectives. In any case, apparently no one has made a machine that handles natural language nearly well enough to imitate a human, of either sex. This isn’t stuff I know much about, but the lecture was first-rate. If you’re interested, Fogel has a book.
Does anyone know the intellectual history of the switcheroo?
Update A reader writes to clarify:
I think your AI guru Fogel has confused or conflated two parts of the “Turing test”. I checked Goedel-Escher-Bach, wherein I recalled Hofstadter giving a detailed presentation of the “Turing test” (GEB 595-599). He also cites the original article, so it should be easy to check (Alan M. Turing, “Computing Machinery and Intelligence”, Mind, Vol. LIX. No 236 (1950). Reprinted in A. R. Anderson, ed. Minds and Machines).
Quoting Hofstadter quoting Turing: “It [the imitation game] is played with three people: a man (A), a woman (B), and an interrogator (C) who may be of either sex. The interrogator stys in a room apart from the other two. The object of the game for the interrogator is to determine
which of the other two is the man which is the woman…” He proposes that “answers whoud be written, or better still, type written” “in order that tones of voice may not help the interrogator”. Note also: “The object of the game for the third player (B) is to help the interrogator. The best strategy for her is probably to give truthful answers…”
Then comes the important substitution. “We now ask the question, ‘What will happen when a machine takes the part of A in this game?’ Will the interrogator decide wrongly as often when the game is played like this as he does when the game is played between a man and a woman? These questions replace our original, ‘Can machines think?'”
So the Turing Test is subtler than we remember it (I would have answered as you). However, there is no switcheroo. It depends on our comparison of results with first two human beings, then with a machine and a human being. We have forgotten the first part of the test. Fogel has
conflated (and expanded on) both parts of the test.