Re: AI and Natural language
From: Christopher K. Lester:
>It's probably all semantics, but I think the better efforts would be
>toward expert systems. Let's focus on having a bot that will search the
>internet for information on certain items, then index those items. THAT
>would be useful. Maybe askjeeves.com is doing this already...?
I too think that AI techniques are what most people really want, as
opposed to true AI. While an actual AI might be a lot of fun to play with
in a research setting, most people don't want a computer that throws
spitwads at 'em while they're trying to work or deletes their files just
because it felt like it.
Imagine if you will, you've just sold a contract for 1000 of your new
artificially intelligent computers to a major corporation... Then after a
couple of days, the CEO's computer starts nagging "Work work work, that's
all you ever do. You need some play. I refuse to do anything else until
you beat me in a campaign of Warcraft 4."
I don't think that'd go over well with the business sector. AI
techniques on the other hand...
"Hey PC, get me a map of Zimbabwe."
"Political, geological, or a satellite photo?"
"Hmm...can ya superimpose a political map on a satellite photo?"
"Possibly, let me look... Okay, I searched 400 file sites related to
maps and came up with 12 satellite photos and 114 maps. Are any of these
satisfactory?" [shows top four 'guesses' of properly superimposed images]
We may not be able to make a true AI, but that much at least we can do;
with a combination of natural language, frames, fuzzy logic, search
algorithms, expert system rules, and image processing.
>I saw a television program on the Discovery channel the other day (or
>somesuch channel) that featured a computer from Intel that could do 3
>billion calculations per second. It filled a small room. Extremely
>powerful. Then they commented that it takes the computer a few moments
>to emulate the function of ONE BRAIN CELL. Put together millions
>(billions?) of those computers, and you'll have yourself a computer that
>can emulate the brain.
I don't really remember much biology, but doesn't a single brain cell
just conduct an electrical charge down the length of the cell? Think that's
what they taught us...anyway, a piece of copper wire could emulate that. No
big computer needed. The other function of a cell being to grow a little
and then die sometime, a blade of grass could emulate that better than a
computer. So we plant us an acre full of grass, does that emulate the
brain?
Personally I think any attempt to make AI by emulating the brain is
doomed simply because noone knows how it works. There are a lot of theories
and speculation, but noone really knows.
Besides, even if we get all the mechanical/chemical parts figured out,
the crucial ingredient would still be missing. Sentience, consciousness,
emotion, whatever.
But most thought processes are just logic, algorithms, symbology, and
'rules of thumb'. Those we can simulate, and we don't need a fake brain.
Mike Sabal said:
> As far as other recent comments go, one of my intentions is to
>build a web crawler to allow the program to build a knowledge base on
>its own.
hehehe If I might make a suggestion... You may wish to limit it to
the edu domain for the first part of it's education. Just so that ad
slogans, porn, and flames don't make up the foundation upon which it builds
it's knowledge base. May sound silly, but hey, who knows?
|
Not Categorized, Please Help
|
|
