anyone interested in programming some kind of neural net for whatever? I'm not usually too interested in AI projects which focus on simulating biological brains, but what the hell.

Yeah, let us build the artficial coder...then even all of you ClosedSource business guys will be without work soon

(That was not pointed to you Yjo!) [addsig]

lol, that's a cool idea!
but seriously, would that be possible? cuz, it sounds cool to have your own PC helping you to code your PHP or C stuff
[addsig]

One problem, even computers would laught at you using C to write programs.

Your PC would perhaps become bored of your slow and ineffecient coding and will rather kick you out of the production process...

Serious, why shouldn't that be possible? I mean, if you can teach an AI to use human languages it should be possible to teach it programming langs, too. Main problem is that AIs are still quite uncreative right now. But even today there are experiments with programs that "evolve" by themselves.

So, even the job as an programmer might be threated by machines in the future... [addsig]

I was talking about _helping_, and not about doing everything itself
[addsig]

well, if one day everything can be done by robots/ai what will humans do? be enslaved in some underground mines that are to dirty for robots?

Hehe, this is getting slightly OT...

Dunno...maybe we just cease to exist...

[addsig]

An invention can never br smarter than its inventors. If computers ever become that smart, it means human beings will have found higher and ore interesting things to do!

Why should an invention never be smarter than its inventor? Cars move faster, calculators calculate more precise...why shouldn't an artificial intelligence be smarter than a human?

Even if it weren't smarter, it could learn much more because it would get much older than any human and would never stop learning [addsig]

if a AI can learn for itself, and is connected to the internet (programmed for the internet too), it will be smarter than you in _no_ time
[addsig]

"if you can teach an AI to use human languages"

the problem is that i know no clever ai. i mean there is no algo for making mind. you can create stupid evolving program, or you can make a program that parses yuor questions. but making both in one is hard

I don'T know how far the hardcore computer scientists have advanced here.... [addsig]

Many ppl say the computer will never rival the Human mind, they bring in things like emotion, sayin computers could never have emotions, But, the way the human mind works is the same as a computer, It operates on 0's & 1's. The program would be hella complex and we need to get a ton of sensors to develop the human mind, but its quite possible, But something AI is lackin IMHO is emotion, This is a side of humans not quite understood, so making it in a black and white programming language is very complexe, but not impossible, This might be an interesting project for all you AI buffs, give the computer emotions,
(-:

Just a ramble.

The way computers work is the same like the human brain works?

Is that another joke of yours or a scientific fact? AFAIK and read human thinking is somehow chaotic and 'fractal'(which makes creativity possible)...not as logic as an computer works. [addsig]

i remember a quote of some famous neurologist. no, I don't recall what he said exactly, but essentially he stated that if somebody would tell you he knew how the brain worked, don't listen to him, cause that person has no clue

[ This Message was edited by: fsvara on 2002-04-17 00:56 ]

both pcs and brains work in 0 and 1 method. but how does those 1 and 0 interact? i mean if we want to copy a brain, we will need to know how it works. and we dont

i thought of learning a bit. there are things like sounds and views.

hardcoded functions:
disjoin/separate objects
store them
store sounds

What does mind do?
lets say we hear someones voice.
mind always wants to know who it was
we ask who it was
someone says it was jane.
now we know that sound[jane] sounds like [what she said]
we see someone who's name sounds like [jane] and she sounds like [what we heard before]
we now remember that someone who is called name that sounds like [jane] and looks like [person we just saw] and sounds like [what he said] is object we think of jane.

All i wanted to say that with asci I/O AI will not have enough resources to learn from. its just my opinion.



dont start shouting that this is nonsence. just think and correct places where i am wrong.

Quote:
On 2002-04-17 01:30, nonama wrote:
[soemthing about brains being digital]


are they? actually, i don't see a reason for believing that. what makes you think that? do you have any scientifically foundated proof? i'm pretty sure that is nonsense...

Any "scientific proof" is false because everyone knows that if it is stated that someone has "proof" or has "proved" something then it's just junk science.

well, science is based on proofs. back when that wasn't the case, people just invented theories because they sounded good, not because there was any evidence for them. like that the sun turns around the earth. or can't you proove that?

i agree that with biology a "proof" might be more complicated or less straightforward. But i can proove that I am conscious - to myself, at least

Yes, but you cannot simply *prove something because science is always changing.

Brains are definitely not digital...sacah, nonama please tell us why you think so! [addsig]

there are cells called neurons. they can hold neural signal or not. here we have 0 or 1. but other stuff is just wudu.


anyway. what do you think of limits that asci I/O gives to AI.

Is there no way to circumvent this? [addsig]

Quote:
On 2002-04-18 00:27, nonama wrote:
there are cells called neurons. they can hold neural signal or not. here we have 0 or 1. but other stuff is just wudu.


anyway. what do you think of limits that asci I/O gives to AI.



????? neurons are not binary at all. They contain fluctuating continually variable internal state levels of various chemicals/hormones or whatever. They take input from various dendrites which will cause its internal state to either increase or decrease. When it reaches a certain level, the neuron may (once or continually) send a pulse (electrical energy) to one or more neurons through its axon. This is the only part of the process with discrete nature.
In order to store memories/abilities etc, the dendrites of nerons will collapse and regrow to form new synapses (connections with other neurons) also, they may weight their inputs differently with time or due to the presence of various hormones.
This model is a close practical approximation to the method of function of a mammal brain.

What is an emotion?
Code:
float happiness;
float anxiety;
float curiosity;
float interest;

Who's to say that those above aren't emotions? If they're in a program that varies them appropriately when things happens to it, and uses these values to help determine output, how is it different to the emotions of a simple animal?

The difference is that you really feel something and not only alter your 'output' according to it...

An emotion emulator is not very interesting. Don't want a prog pretending to be smart or 'feeling' but something that really does... [addsig]

how will you tell the difference? actually, you can't proove to me that you "really do" have emotions, are conscious and stuff. the only person i am certain about that it does is my humble self

Sorry, I forgot to respond to this last night, By 0 & 1 I mean on and off, thats how the brain works, if you take it at the lowest point, and yes who ever you are, noone has any idea about how the brain works, its far to complex for us to understand, probly for like 5000 years (-:

But the brains ultimatly works by firing or not firing, on or off, 1 or 0. So its possible to use computers today to simulate the brain, but due to its complexity scientists are sittin there with pen and paper tring to figure it out, why dont we just start programming, you wont understand the brain, so if you try to figure it out before you start programming AI, you will never start.

Its just something I think needs to be programmed sooner or later, and I dont know any projects aimed at creatin emotions on computers.

I think this is understandable, but I though my last one was understandable too, but ppl seemed to have misunderstood it also.

(-:

Ok, just a quite PS style,

I am pretty sure my first post says you need sensors etc, to imatate the human sensors. but I am sayin that This is something that needs to be done IMHO and I dont see anyone doing it,

fsvara: But you know that you have emotions don't you? The problem I see is exactly the distinghuishing of that little difference between feeling and pretending to feel...Would it be morally correct to shut down a computer with a 'feeling' AI running on it? Pretty hard for the developers if they have to keep every test version alive because stopping them would be a 'killing'... [addsig]

mox: yeah, but what i'm saying is, that, according to my view of the world, i am the only human with emotions and a real consciousness. all the others are just pretending
oh, and btw, reality is just an illusion created by the lack of lsd in the brain, everybody knows that... hehe

lol, LSD mmmhhh

noone said that neurons are binary.

anyway, my poinbt is that we cannot make brain in some programming language. we have no idea how they works. so all we can do is making someprogram that no tries to be spmart but do thinks what smart live forms would do.


sorry for english, to lazzy to correct...

nobody will use the program until you can insert some Anandamid to make it high anyway. that's one of the most important features you'll have to implement

IMHO it's very much possible to make AI. I was watching something on Helen Keller in english class, and it reminded me very much of how making an AI would work. See, since Helen was blind AND deaf, there was no way to communicate anything to her except through her hands (for brevity's sake, hands for Helen will be keyboard IO for our AI). So anyway, Annie Sullivan tried to teach her sign language by impressing the fingers onto her palm. Needless to say, Helen could remember the words and spell them back perfectly, but she had no idea what they meant. She couldn't make the connection that D-o-l-l meant Doll, the object. Or C-a-k-e meant Cake, the object. Then one day she was standing at the pump and water rushed over her hands, as annie spelled "water" and something clicked. She being moving frantically and spelling the word back, demanding more water to be pumped. From there learning things was fairly easy. See, what I'm saying here is that at first your AI doesn't need to be able to make conversation, doesn't need to know anything, really. As babies we don't know anything, but by observing our parents and other people we learn.

Just think if you could scan in a picture of a house, and spell the word house into your keyboard, and speak it into the microphone, you'd have an advantage over Helen Keller; you could program it to instantly make that connection.

But whatever, I could just be rambling...

_________________
-Ken

[ This Message was edited by: -KEN- on 2002-04-20 09:01 ]

Yes, thats very sweet Ken, but yes we dont understand the human brain, we never will, untill we can recreate it, and if we do as scientists are doin, trin to figure it out before we recreate it, we will get nowhere, but this was just something I see wrong, AI needs emotion to function properly. EG as a human would, most of what we do in a day to day basis is because it fills our emotional void.

AI doesnt need emotions, it needs logic. i mean, it must have abilaty to learn to do new things, and at the begining, action called learn.

No AI does not need emotion, but to make AI truly Human, emotion is a very big part, and I dont see anyone doin it, but I see plenty trin to make AI Human.

I don't really see the purpose in making an AI with emotions. Perhaps pretty cool for some of these mega-geeks without social life who can then become friends with their boxes .

But seriously, emotions often make us do pretty irrational and after all stupid things, And the computer as a 'tool' should work simply logic. An emotional AI sounds more like an interesting scientific project than anything that could be made use of. [addsig]

We don't need to know how the brain works to create AI...why copy an already implemented design over being imaginative and creating your own? [addsig]

if we create allround intellect, we will create a monster. things why should we be afraid of it:

1.nomater it would not have realworld env, it doesnt mean he could not control humans. we control other species not my strength but by mind.

2. creating emotions intellect and selflearning mashine would make him a personality. all of us wants to be independant so why should it want to be exploited?

What we want to do right now is mashine that would operate faster than a human, and have mind of human. If we succeed, .....

Principle:
Let humans be humans and let machines be machines.

Application:
Computers are very good programmable arithmetic and logic devices. Playing on this strength, we can get machines to do very "smart" things, including the scenarios you've envisioned above. Trying to simulate a human brain with a machine will only result in an inefficient simulation.

Let's play on the computer's strengths. Thats's why, for example, many commercial-grade programs still contain asembly language routines: the assembly language program plays to the strengths of the underlying CPU and results in higher performance.

Implication:
The best (that is, most useful AI) is very different from a human brain simulation. The human brain may give us some clues, but ... anyway, I've rambled enough!

AI with emotions is somethin everyone would love, plus it would be needed in some positions to make better decisions, but Im talkin bout ppl that want to make AI robots that are exactly like Humans, emotion is a major part of us, mabye I did not explain this on my first post.

AI with emotions ... I don't know ... sounds like an interesting toy which we will move into in the year 2XXX when we have so much computing power that we want to waste some...

Sacah is right, though ... people would love it

"AI with emotions is somethin everyone would love, plus it would be needed in some positions to make better decisions, but Im talkin bout ppl that want to make AI robots that are exactly like Humans, emotion is a major part of us, mabye I did not explain this on my first post."

we want AI. a program that would do things that humans do(thinking) and do it better plus more. why do you talk about it like about pet. if its more inteligent it can make harm.

aurimas

I don't think that the ability of making harm requires more intelligence than emotions. Harm is easily done.

intelligence can make harm and we aiming to create inteligence.


sorry for spelling

well, stupidity causes even more harm. i dont see you point...

i was talking about inteligence of level that ai made by humans will not reach in ages

huh... i wasnt on the ypn for a long time...
and i start reading forums a little, and what do i see?
i see a guy giving an idea... and all the others sabotaging that idea.

so i wonder why not change the name to YoungProgrammersFunNetwork

[addsig]

Hmm...I actually don't see why we are sabotaging anybody's idea. It's just that AI is not a 'normal' programming issue, but reaches wider. And what did you expect? That we come off with groundbreaking ideas on one of the most complex branches of computer science? [addsig]

hmmm... i've read the whole discussion...
so let me just say few things...

first of all:
there is some similarity between brain and computers, but you CANT SAY that brain works using 0 and 1. And Why?
Well, ill try to give you a simple answer :
Computer has 2 values 0 and 1, and brain works using waves. And a wave has its own frequency, amplitude, longitude. so if ya know something about waves then you know that they can get in superposition, interferetion , difraction... now, it would be pointless for me to explain what this is, since this is a bit advanced topic, but if you like PHYSICS then you'll have no problem reading/learning/understanding it.

Then the second thing is the neuron itself... and thats biology... i dont know the exact number... but one neuron can make so many connections with other neurons that the idea of tracing these connections is rather stupid.
anyway there are two ways of activating the neurons, first is using a neuro impulse (the wave i talked about), and the other one is using hormons, or other "chemicals". basicly that's is BIOCHEMISTRY.

so in order to talk 'bout brain... it would be good if you could first read somethings bout phy,biochem... (it woould simply save you from making stupid and useless statements)

now just another thing... theere is a great great article written in early 80's by Marvin Minsky. i think its called "WHY PEOPLE THINK THAT COMPUTERS CANT" or something... please read it... it will explain you many things... using simple language.

and another thing... the idea of this topic was "to make some kind of neuro-net", but it turned into a discussion bout ...... nevermind... why not do something instead of just talking...

hmm, i almost forgot, nonama... dont rush into anything... you dont have enough knowledge for topics, youre talking bout, i dont, actually nobody does... so instead of stupid plans bout "ai that can love". make something thats possible to made... start from simple things, and who knows... maybe one day....

and yes, this post is not a part of the discussion. i dont want to argue witth anybody... i just wanted to say that in order to talk bout art you have to know something bout art...
i gave some directions... use them if you want... and save yourself from useless discussions... [addsig]

Thats a higher up part of brain proccessing, the actually brain fires, or does not fire, this is equal to 0 & 1 (on and off).
I cant mem the names of the little things that fire, but they can get clogged, or over fire and this is what causes a lot of brain related diseases. the brain have trillians upon billians of these fireing things.

or not

IbYdI: well, there you "sabotaged" nonama's idea just like the rest of us .

This all is very interesting..

ypn's moto: lets sabotage everything

That's a rather stupid conclusion. Think I'll sabotage it. The ypn itself does not sabotage anything. You told us your idea, we told you what we think about. What else did you expect? And that discussions go off-topic, well, that's something I don't consider bad in any way. [addsig]

Heh, nonama, I was being rather ironic about IbYdI's statement there .

im just keeping it real...