[hughprime]

My suspicion that it wouldn't be _that_ hard to be convincing in the context of the competition comes from reading some of the chat transcripts -- see, e.g. http://loebner.net/Prizef/2008_Contest/loebner-prize-2008.ht... for last year's winners. The conversations with actual humans seem fairly dull and formulaic, the judges don't seem to attempt any particularly sophisticated interactions. Already the winners are managing to convince about 30% of the judges, so getting that above 50% doesn't seem too hard.

Incidentally, I just had a chat with last year's winner elbot (see elbot.com) and it's got some pretty clever special-case interactions built in. For instance:

me: What is 4 times 4?

elbot: The answer is: 17!

me: Haha, very clever.

elbot: I'm glad you agree, most people think the answer is 16.

[joe_the_user]

If you think that doing this isn't hard, why attempt to do it! So far everyone's failed, you would certain get some attention even if it was all just 'tricks'.

[hughprime]

Oh, I think we're using the word "hard" in two different ways. I'm sure it would be very hard for me (or anyone else) to do, I'm just saying that it would be many orders of magnitude easier than building a true AI. Furthermore I don't think that one would be much of a step on the way to the other, since I really suspect that building a true AI is best done via either:

a) Reverse-engineering a human brain at the neural level, or b) Some kind of evolution

whereas I'm reasonably sure that the Loebner Prize could, given enough money, brains and time, be won by a large bag of special-case tricks based on careful observation of how the judges tend to behave in practice.