[net.followup] Refutation of "Refutiation of the Turing Test"

custead (11/25/82)

	....  If you were given a book in English which described
	how to combine the Chinese characters, and if the book
	was good enough, eventually you might get good enough
	at manipulating the formal symbols of Chinese to fool
	even a native Chinese speaker into thinking you knew
	Chinese, but you really don't.

There are 2 fuzzy qualifications in the above:
   "if the book was good enough", and
   "eventually you might get good enough"
YES...eventually you would get good enough at manipulating
the symbols of Chinese to *fool* even a native Chinese speaker...
this point would come when you had learned the rules of Chinese...
To learn to manipulate these symbols that well you would come
to learn the language...starting from scratch the way a child does...
although "if the book was good enough" it could be a lot of help.
So the last line... "but you really don't" is wrong...you really DO.
You are given a series of symbols and they now are meaningful to you
and you are able to give a meaningful reply.  I myself can
manipulate the symbols of English well enough to fool a native
English speaker, but it has taken me years.  By the way i do
not know english and you cannot prove that I do.  I am merely
manipulating these symbols inside my head in response to some
symbols that I saw on the net.

If a computer is able to do
this equally well, even if it "doesn't know English", I will
say that it has passed the Turing Test.  And as long as it
can carry on an intelligent conversation (via symbol manipulation)
it does not matter if you want to say "but it really doesn't KNOW
English."

				custead
				of the western provinces