Re: [MLUG - DISCUSSION] Re: [MLUG] AI - any suggestions?

[Michael <EMAIL:PROTECTED>]

Mood swings suck monkey ass. Damn I could at least be pregnant if I must
suffer them. Nine months from now can't I give birth to a laptop or
something? :P

> i resent being refered to as "people." please don't generalize.
> arguments i have presented to you are not pulled out of pop sci
> magazines, but are in fact present in any intro level AI textbook. 

My opinion of current AI textbooks is that most are little more than pop
sci. There are exceptions but a lot of them really suck. There is a lot of
tucking of tails and running.. people who made big claims 20 years ago to
get grant money and now are embarassed to have done nothing more than
invented washing machines that get your cloths much cleaner. Have some
pride in your accomplishments, explain how you did it, and give some
theory on where things might be going.. AI is still in many ways a new
frontier.

> it is extremely difficult to show code that disproves a statement like
> "parsing nlp is easy." fortunately the burden of proof is on you. if you
> don't want to get "sick of arguing" in the future, perhaps you can
> refrain from making statements of this type, so that you won't be asked
> to prove them.

Well if you're going to say that XYZ doesn't work then give a bit of code
explaining why you think it won't. It's equally hard to refute doubts when
all you get is superstition without any evidence. If you think you have a
better way then show some code for that or specs or something. I don't
mind discussing things but when everyone is arguing against me and I go
from having to explain one line with hundreds it is annoying. Every level
of explanation I gave people wanted to take it to the next level of
argument and then were upset that I was posting to much.

> i made my original reply because i was interested in finding out the
> details of your work and how you tackled specific problems, and hoped
> it'd come out in the discussion. from the tone this exchange has taken
> it looks as if it's heading down in flames - maybe we can resolve it
> some other time over a pitcher or two of some stimulating beverage

I explained the basic theory of parsing language I tend to use. With each
project I tweak my methods and make them better. Sometimes they are
implemented as code and sometimes by other methods such as neural nets..
or a combination thereof. I fully agree with everyone that it takes a LOT
of work to make a True AI but the basics of breaking English down into a
usable form aren't very difficult.

If you want more explanation of my study into true AI I've done a lot I
find interesting with building psuedo-emotions into psuedo-intelligent
systems. I also have a breif blurb on my projects page about some basic
image processing stuff I'm tying into my image gallery software as I feel
to comprehend language you must comprehend the meaning of the words
meaning the AI's "brain" must have large datasets from all sensory methods
used in human speach. Of course this also has the somewhat interesting
angle that my software is very similar to images.google.com except works
based on user-moderation.. by tying a reasonably simple AI into the system
unknown images can be placed in categories by the AI before users have
searched them.. but the AI would be a normal user so it could have it's
choices changed by users.. providing feedback education.. and it
increasingly makes my software more exact. Much more exact than any other
search technology I've seen. I haven't decided if it makes sense to
release the AI's code yet though as it'll only be meant to be an addition
to the already open system I'm working on.

--
To unsubscribe, go to http://mlug.missouri.edu/members/edit.php

Archives are available at http://mlug.missouri.edu/list-archives/