2
$\begingroup$

Did Alan Turing expect the AIs to be aware they were being Turing tested while the game was being played?

I think it's slightly harder to look like a duck while pretending to be a duck than when you believe you are a duck. It is a step further.

Aside from all other criticisms of original article, did Alan Turing have the time and motivation to address the point in any later writing?

$\endgroup$
  • $\begingroup$ Hi this has been mulling around in my head, is a consequence of your question whether a human imitating another human would they fail the Turing test? Plus are you saying an AI capable of passing the Turing test would believe itself to be human misgiving physical differences if it was unable to observe a human. If a simulation convinces the reciever it's not a simulation isn't that simulation no longer and is real, because you can't distinguish reality from that simulation. But I guess it also needs a measure of determining reality in order to judge the comparison. $\endgroup$ – Bobs Dec 12 '17 at 22:33
1
$\begingroup$

First, I think it is important to mention that the Turing Test as is currently accepted is an updated version of Alan Turing's proposed imitation game, so your question is twofold. I don't think Turing makes this distinction that you propose. To Turing the question was "Can machines think?" and he explores this by way of the imitation game.

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 front the other two. The object of the game for the interrogator is to determine which of the other two is the man and which is the woman. He knows them by labels X and Y, and at the end of the game he says either "X is A and Y is B" or "X is B and Y is A." The interrogator is allowed to put questions to A and B

It is A's object in the game to try and cause C to make the wrong identification

The object of the game for the third player (B) is to help the interrogator.

This is Turing's outline of the imitation game (shortened) between a man and a woman. The only difference between this game and the one we care about is that we replace party A with a machine and the only requirement on B is that they are human.

So yes, in Turing's imitation game the machine is privy to the information that it is not actually human. Whether the machine uses this information is up to the machine.

To talk about your statement on pretending to be a duck versus believing you are a duck lets examine the base imitation game with ducks and frogs. Let's say player A is a frog (either pretending to be a duck or truly believing that it is a duck) and player B is a duck. Suppose for this example, the frogs and ducks are able to speak English and follow the rules of the game. The interrogator may ask "what is the color of your beak?" The duck will respond yellow, assuming its beak is yellow (I am no bird expert), and then our smart frog pretending to be a duck will look up the color of the beaks of ducks and respond yellow as well.

However, our frog that believes he is a duck will think "I am a duck so to discern the answer I must only look at my beak." And thus, our frog will say "I do not have a beak, but my mouth is green." This will oust the frog.

$\endgroup$
  • $\begingroup$ The frog would actually be delusional and report its beak as yellow. $\endgroup$ – bruno Jan 22 at 20:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.