Maybe not seperatly, as they all tie in of course, but Ive seen so many
projects now that try to do the reasoning aspects of AGI, and they all seem
limited by one main factor, a lack of knowledge, or base facts in the system,
and whatever facts are in the system are hand-engineered and few. This seems
to make good demo's but fragile systems if we want any kind of expansion.
It gives a godo test bed for learnign about how the systems should work,
but we dont have the ability to hand engineer all the many facts, and try to
make them all work together.
Russell,
Its a catch 22 really,
>"In other words, any successful NLP must incorporate the same knowledge of
the universe that the speaker possesses. Without that knowledge, natural
language cannot be understood."
We cant understand english because we dont have enough knowledge, we dont
have enough knowledge because we cant understand english.
Something in the line of thought must be wrong, and we CAN communicate with
computers now, albeit a limited form of english in simple commands, so we need
to up it, and give it the ability to know more and more english, as the basis
for a NL interface to teach it any knowledge it needs on an interactive basis.
And the above quote is not strictly true on an obvious level anyway, no one
person knows everything another person does, or they could never learn.
Given a basic AI, that knew some words and info about food and eating
"I feasted on an olive-and-pimiento-cheese sandwich on wheat bread."
The AGI could look at that, say I dont know what "feasted on" means, and I
could tell it that it is very similar to "ate" and it would learn that small
bit of info.
So I talked with it, using words it had no knowledge of, and it learned from
me.
The trick here, is what is the base smallest amount of words and knowledge
needed to gather most other knowledge, and what is the easiest interface and
reaction for the AGI, to learn these in a good fashion.
>I think we can all agree that giving a computer the same knowledge of the
universe that a >typical human possesses is a task beyond even our hubris at
present. Until we can solve >this problem, we have to dismiss the possibility
of natural language user interfaces."
So I think that shows pretty simply that all the worlds knowledge is not
necessary to make an AGI, bu tthat learning and teaching must be used ot get as
much of that knowledge as possible... To date, this is the only real way I see
of getting a complete decent KB to an AGI, though text extraction and some
amount of engineering low level information must be used as well. And any other
DB's and ontology can be used to assist these tasks.
James Ratcliff
> Learning the English language (or whatever language) seems to really
be the
> blocking point for most every AGI project I have heard of, and I
would like
> to see a lot more interst and work done on that, even possibly before
the
> overall structure of the rest of the AGI is concerned with, or at
least a
> much mroe significant portion.
Ability to learn English is one key requirement of an AGI
architecture, but I don't think it can be considered separately from
the rest of the architecture...
Ben
Russell Wallace <[EMAIL PROTECTED]> wrote:
On 4/20/07, James Ratcliff <[EMAIL PROTECTED]> wrote: Learning the English
language (or whatever language) seems to really be the blocking point for most
every AGI project I have heard of, and I would like to see a lot more interst
and work done on that, even possibly before the overall structure of the rest
of the AGI is concerned with, or at least a much mroe significant portion.
I just found an article by Chris Crawford that I remembered reading years ago,
with a relevant quote:
"In other words, any successful natural language processor must incorporate the
same knowledge of the universe that the speaker possesses. Without that
knowledge, natural language cannot be understood. I think we can all agree that
giving a computer the same knowledge of the universe that a typical human
possesses is a task beyond even our hubris at present. Until we can solve this
problem, we have to dismiss the possibility of natural language user
interfaces."
http://www.erasmatazz.com/library/JCGD_Volume_9/Little_Languages.html
_______________________________________
James Ratcliff - http://falazar.com
Looking for something...
---------------------------------
Ahhh...imagining that irresistible "new car" smell?
Check outnew cars at Yahoo! Autos.
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=231415&user_secret=fabd7936
