Ben Goertzel wrote:
Well, having an intuitive understanding of human language will be useful
for an AGI even if its architecture is profoundly nonhumanlike. And,
human language is intended to be interpreted based on social,
spatiotemporal experience. So the easiest way to make an AGI grok human
language is very likely to give it an embodiment in a world somewhat
like the one in which we live. This does not imply the AGI must think
in a thoroughly humanlike manner.
-- Ben G
Call me old-fashioned, but I believe "..an intuitive understanding of human
language..." can be acquired entirely through description. Why can't human
language be "...intended to be interpreted based on..." *description of* social,
spatiotemporal experience? Of course it's not the "same thing." But is that a
distinction that makes a significant difference? IMHO, the answer is: not if
all you really need is an AGI with *human compatibility*. I submit we can build
a highly-functional, human-compatible, AGI without requiring that it directly
experience human space and time. A description should suffice.
You've written books. In those books, I assume you described all manner and
sorts of things. In order for people to learn from your books, are you really
suggesting that they have to directly experience everything described therein?
If so, why not skip the book-writing drudgery and lead learn-by-doing seminars?
You'd make more money and think of the frequent flier miles you'd rack up! :-)
I'm not saying that learning human language by direct experience couldn't be
beneficial. Nor am I arguing that it is not important for an AGI to be able to
communicate effectively with humans. Understanding human language is probably
the "gold standard" for this behavior and it certainly will be more important
for some applications (e.g., humanoid robots) than for others (e.g., data-mining
or security AGIs). My issue is with how Turing sent everyone out on this wild
goose chase fifty-some years ago and many AI/AGI researchers are still "out
there" looking for Turing's if-a-human-cannot-tell-it's-not-human goose. I just
don't think "Turing-indistinguishably" is necessary. We should be concentrating
on building artificial general intelligence (AGI) that is human-compatible, not
artificial general human intelligence (AGHI). I submit if we build the
human-compatible AGI first, it will help us achieve AGHI in much less time and
at much lower cost -- if that's what human society decides it wants.
Research AGHI? Maybe, but low priority. Research and implement
human-compatible AGI? You betcha. And, toward that latter task, I'll put my
money on the learn-by-description approach until direct experience shows me
otherwise! :-) <- That was (an attempt at) a JOKE.
Brad
Grounding is a potential problem IFF your AGI is, actually, an AGHI,
where the H stands for Human. There's nothing wrong with borrowing
the good features of human intelligence, but an uncritical aping of
all aspects of human intelligence just because we think highly of
ourselves is doomed. At least I hope it is. Frankly, the
possibility of an AGHI scares the crap out of me. Personally, I'm
in this to build and AGI that is about as far from a human copy
(with or without improvements) as possible. Better, faster, less
prone to breakdown. And, eventually, a whole lot smarter.
We don't need no stinkin' grounding.
Cheers,
Brad
-------------------------------------------
agi
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/
Modify Your Subscription: https://www.listbox.com/member/?&;
<https://www.listbox.com/member/?&;>
Powered by Listbox: http://www.listbox.com
--
Ben Goertzel, PhD
CEO, Novamente LLC and Biomind LLC
Director of Research, SIAI
[EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>
"Nothing will ever be attempted if all possible objections must be first
overcome " - Dr Samuel Johnson
------------------------------------------------------------------------
*agi* | Archives <https://www.listbox.com/member/archive/303/=now>
<https://www.listbox.com/member/archive/rss/303/> | Modify
<https://www.listbox.com/member/?&;>
Your Subscription [Powered by Listbox] <http://www.listbox.com>
-------------------------------------------
agi
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/
Modify Your Subscription:
https://www.listbox.com/member/?member_id=8660244&id_secret=108809214-a0d121
Powered by Listbox: http://www.listbox.com
