portegys (11/28/82)
In reference to the recent note commenting on the counterexample to
the Turing test.
That note really hit me strangely, and after a while I figured out why.
I think that Searle's criticism of the Turing test is based on what he
sees as a weakness, but it is exactly that weakness that is its
strength. By that I mean that Turing decided to sidestep all of the
subjective and intuitive definitions of intelligence and become a
behaviorist for a while. And in doing so he proposed a method of
determining (defining?) intelligence based completely on external
measurements. So if there is no way to tell the difference between
someone who "knows" chinese, and someone who can always respond in a
way that can fool people, is there any sense in making a distinction?
Is there any distinction?
Tom Portegys, BTL IH, (...ihuxv!portegys)avsdS:avsdT:wcl (12/14/82)
Yes, there is a difference between someone who "knows" Chinese, and someone who can "merely" give the right answers all the time. It is subtle, and possibly not rigourously definable, but it is the same as the difference between someone who's really "all there", and a psychotic who's only faking normalcy. If you talk to someone who's distanced from reality for long enough, you'll know the difference. Most of the AI simulations of intelligence come across like someone who's either bored, but far too polite to let you know, or who's very single-minded and refuses to talk about anything else. Anyone in "real life" who NEVER displayed the ability to talk about anything other than medical diagnosis (or oil drilling, take your pick of the "expert" programs) would be classified as not sane. The relation between sanity and intellignce get's VERY hairy, so I'll bow out at this point... Bill Lindemann fortune!atd!avsdT:wcl