Suppose that a box was cleverly carved so that it looked like it had a towel
draped over it. A visual based AGI program would be unable to detect the
difference without some kind of additional action to help it discover the
trompe l'oeil. And suppose that a word was used to refer to different things.
A visual based AGI program would have the same kinds of problems understanding
that as a word-based AGI would have unless some kind of education to point out
that the word was being used in different ways was available to it. An AGI
program has to be able to effectively utilize education. It has to be able to
meaningfully convert instruction into workable knowledge. The distinction
between procedural knowledge and declarative knowledge for a person is not that
distinct except when looked at in detail. (The decision to call certain mental
events "procedural" would be somewhat arbitrary.) The ability to be educated is
one of the hallmarks of intelligence. It should not be disregarded. And this
can be achieved in text-based AGI. It is just a matter of when it is done.
Watson may have been long overdue but it was a major milestone in AI/AGI. Jim
Bromer
From: [email protected]
To: [email protected]
Subject: Re: [agi] What I Was Trying to Say.
Date: Fri, 10 May 2013 14:59:02 +0100
I’ll gladly put $1000 (or considerably more) down now publicly that neither
your nor any other word-based “so-called AGI” prog will generate a single thing
in 1/2/5 years – generativity, I think we can agree, being a test of AGI.
-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424
Modify Your Subscription:
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657
Powered by Listbox: http://www.listbox.com
