From: Harvey Newstrom (mail@HarveyNewstrom.com)
Date: Fri Jun 29 2001 - 20:38:26 MDT
Robert J. Bradbury wrote,
> Aha, but are you asserting that search engines, databases,
> lookup tables and fuzzy logic (I'll remove chatbots) cannot
> emulate a significant fraction of human behavior?
Not at all. These techniques can easily be incorporated into AI software.
However, GAC does not seem to be AI software.
GAC is like a calculator that can't do math. If you type in a bunch of math
problems with answers, it can answer those problems you already typed in.
But it can't answer new math problems, because it doesn't know how to do
math.
Similarly, GAC reads in millions of Usenet posts. When you ask it a
question, it searches through this database and spits out pre-recorded
statements from these Usenet posts. It can't create new conclusions that
aren't pre-loaded into its memory.
All that talk about neural nets were future possibilities where this
database could be fed to a neural net to see what happens. GAC does not
currently have a neural net. It has a database.
> If most other human beings did *not* display an extraordinary
> amount of behavior that can be specified by a finite state
> automata it would *not* be safe to go out in the world.
>
> Human interactions rest on the assumptions that much of
> the behavior of fellow humans can be predetermined.
> If humans can predict it, then I see no reason why
> an "artificial intelligence" cannot be designed on
> the basis of relatively simple methods that should
> expect the norm (bus passengers should sit quietly)
> or produce them (I a human-like AI must sit quietly).
Nobody is saying that A.I. can't be predictable. A chess program is
predictable, but it simulates chess playing. A calculator is predictable,
but it can solve new problems. GAC seems to be only to answer questions
with pre-recorded answers. It is more like a tape-recorder or a floppy
diskette. If you record data on it, there it is. That's data retrieval.
It's not even data processing.
-- Harvey Newstrom <http://HarveyNewstrom.com> <http://Newstaff.com>
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:08:22 MST