Artificial Inteligence

You know..Good question. But I would presume that the artificial would use a language far more advanced and efficient than we could comprehend. For we as humans, we use our language(s) for the purpose of communication but, it really depends on the area. I.E...People from the U.S. talk faster than most people in the world because we want things fast and efficiently but, people in...Austrailia talk slower than most (this is just an in-acuarate example) becausit's hot and it takes alot of energy to move/speak. A.I would have a language that suits them and their needs. "Phew" I hope you understand what I mean, I can be a bit abstract sometimes.

You know..Good question. But I would presume that the artificial would use a language far more advanced and efficient than we could comprehend. For we as humans, we use our language(s) for the purpose of communication but, it really depends on the area. I.E...People from the U.S. talk faster than most people in the world because we want things fast and efficiently but, people in...Austrailia talk slower than most (this is just an in-acuarate example) becausit's hot and it takes alot of energy to move/speak. A.I would have a language that suits them and their needs. "Phew" I hope you understand what I mean, I can be a bit abstract sometimes. ;)

Well, I'll try to answer the question in the beginnning of the Thread:

O worked with LPA Prolog this year around A.I. and the truth is that I was very suprised, a Logical programming language with backtracking. I was able to do small programs able do answer questions based on a knowledge base given by the user. It could understand (not really understand them of course) my sentences and correct them. Then a simple game with meta-programming (the language defines itself), of course with lots of limitations. The next step was a program do help you make decisions based in factors of confidence, for example, my program should be abel to tell you with some certainty if I should invest on certain stock or not (based on variables that I would give to it).
Well, we can't call it real A.I. but I think we're getting there. The games were mentioned in this thread, it's the beginning of A.I. I remember to play some games that the computer would learn your skills and would adapt to them.

I agree acidreign, we are programmed with the most basic instincts to survive, like to eat when we're hungry, sleep when we're tired, and avoid things that cause us pain. These are some basic rules that don't seem all that hard to put into a logical form a computer could understand. As far as seeing people smiling as a child, we know it means they're happy based on the fact that we do the same thing when we're happy. If we frowned when we were happy we wouldn't know the difference really, that's something learned from a bit of experience combined with the thought that you do the same thing when you're happy, etc...

I suppose there's a need for emotions and pleasure though, it's a tool to help learn, without pleasure/pain we would have no incentive to do certain things, same with emotions, we feel bad when we're let down or someone yells at us so we learn from it, if we don't like feeling that way we do something to change it. I think I was wrong earlier, in order to learn intelligently we need some sort of feedback aside from the external world to make us want to do one thing and not another. Without some sort of emotions or pain/pleasure system we would have to base all our decisions on things we've been told, eliminating the real ability to decide for ourselves.

On a side note for anyone who wants to argue if we have free will, while without feelings as such we'd be confined to believing whatever we were told in most cases, is it possible that emotions cause the illusion of having a choice when in fact we were bound to make that decision regardless? Just a thought.

I wrote a paper on A.I. in uni.

Hii guys'
I have Read all ur post and also i am very much interseted in AI bur i call this as NI natural intellegence coz a machine with so much complex input output system and database capablities is a machine indeep coz i cana't think of its own it we r to make a thinking machine we should make the think for itself and asks the ? to itself only and tries to find the loghival asnwer either by the database or some other logic
there the parts needed for a NI machne is
1. Highly fexible Input Output System whcih can interact not only with the vitrual world but also with the real world
2. Higlhly fexible database like the nurons of out brain which can change the data and append as and when and where needed
3. A Logic to Interpret the data stored into some logical anwer's either in visual for or in text form, i think the visual for will be much better coz by this we can make out machine visualize the concepts and answer's the reason is same a new born child cannot talk he can only viualize and put the ? on himself and seeks the answer from himslef only
4. A logic so that our machine can think by itself and aks the ? from itself only
there for 1+2+3+4= A NI MACHINE

In my opinion the answer to this question is not to use a programming language.

I would use neural networks, although to mimic human levels of intelligence they would be huge.

I would try and design aggreagations of neworks to mimic the areas of the brain we know perfom specific functions which when aggregated together would create the AI.

Several neural networks could form a visual processing centre:
Colour network
B& White network
Vertical Line Network
Horiz. Line Network
Repeating Pattern Network
Vertical Movement network
etc.
etc.

As you can see these networks would require a huge amount of resource to build each component, train the components, connect them together and train the whole.

Of course to test & train the visual processing network aggregation it would need to be connected to a suitably trained aggregation performing visual memory etc., etc..

Not a small task.

Now to get back to the question - You could write a neural network simulator in any language you want but you need one massive Beowolf cluster to run it all on.

But there may be a better solution:

Using a mixture of neural networking and a variation of genetic algorithms so that each neural netork has selection pressures for its survival to the next generation (an also selection pressures on agregations) it may be possible to evolve the AI by simulating a challenging environment for your developing AI.

And then the Terminator scenario starts.

Anyone remeber a sci-fi book called the Two Faces of Tommorrow about training an AI on a space station?

I personally think not only that AI is possible, I think it exists already. The internet. It is massive, it "knows" just about everything about us. the internet is evolving, constantly bending and shaping itself to our immediate needs. I think that is, in a sense, intelligence.

I will use my l33t code to write a super virus to gather information and eventually become self aware.

Does anyone have any spare floppy disks because I'm going to use these as brain cells. LOL

MIT got a nice AI lab. I got buddy who used to study in there. Pretty nice. Life forms + computer generated structures to make a program fro simulating it.

to
mark_boyle2002

I've got a bunch of 5 1/4 disks around

Excellent. All we need now are 2 web cams for eyes, A couple of microphones for ears and a carrot for a nose.

don't forget the corcobb pipe and top hat, we can call him frosty!

A carrot for a nose? How will it smell?

*badum*

Ah, back from leave... it was good... too short though...

Good Gracious. I never thought of that.

All the same I'm sure if we keep posting here this forum is going to become self aware.

Excellent... cept I don't have any carrots for it :\

I think there is a substitute carrot in my wifes top drawer somewhere its a different colour though !!!!

"We need to get the computer to always think of any object in terms of its links, not as a single unit"

What if the links are not logical (in any sense of the word)? For an up to date discussion of machine consciouness by one of the original neural network gurus see "I, Computer" by Igor Aleksander, NewScientist, 19 July 2003, pp.40-43.

Hmmm,

Then we could get it to think of other objects in a different way. in terms or other bodily functions. Like viewing a beautiful woman for example. (Morning Goddess)

That doesn't stimulate my neural network but it does send signals to other areas.

Humans aren't logical anyway. Take a look at mark_b as an example :P

The links will be as logical as interaction allows them to be.

I never though I would convince you wrayth

Logic sucks. Chaos will always be superior.

I am living walking proof !!!!!

All hail Kaos, and may his Dairy produce only the finest milk and cheese.

Anyway, that's the beauty of my AI concept. It will be as logical or illogical as the influences upon it.

So I would be like a logic bomb then. Woot. I always wanted to be a l33t haxoring logic bomb

Making logic seem fuzzy is not as hard as most people think..

I have some experience in writing lame ass apps (just under 1kb) that draw intricate shapes (not fractals but similar)

that way you could also make the logic of the AI seem less logic ;)

All you need to do is let the AI talk to random people on the web. Illogical enough for anyone... just look at these forums :P

Dammit Jinx, your signature hurts my eyes.

set it loose on www.christianity.com that will make it a master of fuzzy logic... no, on second thought, I'd better not, we are trying to create INTELLIGENCE after all.

let it read thrue my collection of html-ed oreilly books ;)
it will be able to make itself a little better.. and hey isn't that what we realy want our AI to do ??

If the AI would be made to be able to program new versions of itself, then you could have a nice little evolution thingy going..

"good" new versions write newer versions
"bad" versions don't function and won't write new versions
"evil" versions make more evil (eviller ??) versions and killoff the "good" versions :D

that would be real AI if you ask me..

Quote:

---

Originally posted here by the_JinX

"good" new versions write newer versions
"bad" versions don't function and won't write new versions
"evil" versions make more evil (eviller ??) versions and killoff the "good" versions :D

that would be real AI if you ask me..

---

Have a look at this - Is this the thing you were thinking about.

http://www.aic.nrl.navy.mil/galist/

So were all agreed ints dangerous to create artificial life forms and intelligence systems because of the subliminal messages in T3 then ?

It's safe to create artificial inteligence as long as we don't allow the AI to watch movies like Matrix, T1-T3 etc..

Better keep it watching mary poppins and not let it near the news.

We can let it watch Space Oddessey 2001. It'll take so long to try and figure out what the hell is going on, it'll never have time to attack humanity

We could make it watch our l33t pron collections and see what it makes of that.

K, quit fixating on the pr0n, mkay? You're worrying us here. We'll have to find a passing AI to shoot you if you're not careful.

LOL,

But I have my sony AIbo to protect me. And Johnny 5 and my ed209 from robocop.

HAHA I am invincible.

Hah, I have a cunning plan involving duct tape that shall laugh at your defences. Yes, I said Duct Tape! Ph34r my l33t duct T4p3z0r Sk1llz!

haha, but I have sticky back plastic and you shall not win Artificial Intelligence because I will fly to your mother ship orbiting high above the earth which conveniently will be running an operating system compatible with the virus I have written.

I am building a baracade of old p75s as we speak

Oh no, I forgot about your l33t skillz that enable you to build a virus complete with cool laughing angular skull effect in one day for my alien operating system!

Still, I respond with my flying squirrel style hopping toad stance!

Damn,

The "flying squirrel style hopping toad stance"

well in that case take this "c:\winnt\setup.exe" ha. you didn't expect that did you
especially not when coupled with the deadly vibrating bum faced goat android

Bother.

Very well then, I shall have to build a mysterious and unseen time machine that moves me back in time with a cool glowing circular special effect with lightning (that somehow only allows organic matter, so I have an excuse to walk around naked and beat up bikers for their clothes) at which point I will track down the developer responsible for winnt and attempt to assasinate him. Failing because of some kind of missplaced moral code, I will then find the factory responsible and blow it up, since my moral code doesn't stretch to covering people who are now jobless and ruined due to my actions, and apparently ignoring any possible paradoxes that might rip apart time and space.

Hah,

Then While you are traveling back in time in your unseen time machine I will travel back in the deloreon which the old guy at the end of my street modified (Works with clothes to).
I will speed up to 80mph and run you over.

I will then take this little blood test machine and a simcording(See Sixth Day) of the winnt kernel guy (Going to do some research before I go) which I will keep backed up in an air duct in my room beside my cigarrettes and jazz mags and the FBI and Secret service won't even look.

So there