Ai Forums Home Welcome Guest    Wednesday, April 26, 2017
Ai Site > Ai Forums > Language Mind and Consciousness > About artificial general intelligence Last PostsLoginRegisterWhy Register
Topic: About artificial general intelligence

fasisi
posted 5/27/2010  09:13Send e-mail to userReply with quote
Hello all,

I have got some idea/question when thinking about artificial general intelligence. I have noticed that human (or any living creature that has brain) has the ability to learn about anything. (Doesn't mean that a spider can learn calculus). Even if we fail on trying to understand something (like quantum theory or number theory, or a spider looking at database diagram) we don't hang (like a computer does). We simply fail to grasp the meaning or simply ignore it.

From the fact that we can learn about anything, I believe that there must be some kind of universal algorithm in the brain. And this algorithm is very flexible. Meaning that the algorithm can build a new algorithm. (That is when we use current knowledge to understand a new one). And also mean that the algorithm can be inside another algorithm. (That is when we use new knowledge to enhanced old one).

I maybe expressing it incorrectly. But I hope you get the idea/points.

Thank you,
fasisi


tkorrovi
posted 5/27/2010  15:26Send e-mail to userReply with quote
Please choose the terms carefully. It has been agreed in some places that Artificial General Intelligence is a stricter term, which means modelling the whole human intelligence, while Artificial Consciousness is a wider term with less strict requirements, ie not all aspects of human consciousness have to be modelled and not necessarily fully modelled. It may seem to be not so important, but in some cases mixing the terms can cause a great confusion.

> I believe that there must be some kind of universal algorithm in the brain. And this algorithm is very flexible.

I think you are right in both. Brain is most likely a fully self-developing system, and the only other such systems, such as cellular automata, are implemented as an algorithm which implements the basic mechanism, and the wholw system runs on that basic mechanism. There is no known other way for making such systems. An algorithm generating other algoritms, well, all algorithmic languages are just so redundant that a vast majority of randomly generated code doesn't do anything, so self-developing mechanism such as in cellular automaton is the only reasonable alternative. And you are very right that the main requirement for such system is that it has to be very flexible.

 Artificial Consciousness ADS-AC project

fasisi
posted 5/29/2010  04:55Send e-mail to userReply with quote
 
tkorrovi wrote @ 5/27/2010 3:26:00 PM:
Please choose the terms carefully. It has been agreed in some places that Artificial General Intelligence is a stricter term, which means modelling the whole human intelligence, while Artificial Consciousness is a wider term with less strict requirements, ie not all aspects of human consciousness have to be modelled and not necessarily fully modelled. It may seem to be not so important, but in some cases mixing the terms can cause a great confusion.

> I believe that there must be some kind of universal algorithm in the brain. And this algorithm is very flexible.

I think you are right in both. Brain is most likely a fully self-developing system, and the only other such systems, such as cellular automata, are implemented as an algorithm which implements the basic mechanism, and the wholw system runs on that basic mechanism. There is no known other way for making such systems. An algorithm generating other algoritms, well, all algorithmic languages are just so redundant that a vast majority of randomly generated code doesn't do anything, so self-developing mechanism such as in cellular automaton is the only reasonable alternative. And you are very right that the main requirement for such system is that it has to be very flexible.

 
Hello tkorrovi,

Thanks for your reply.

Also thank you for the warning about using terms. I am new to this field. I think what I meant was Artificial Consciousness.

I agree with you that cellular automata can be used to build self-developing system like the brain. I must add that understanding complex system is also important in building self-developing system.

I am imagining that the cellular automata is running a basic mechanism which behave differently if connected in a network.

Is there really no known way for making self-developing system? Because I have read that the IBM have build a computer or system that mimic the rat brain. And other laboratory have claimed that they have developed system very similar to the human brain.

(Check this: http://hplusmagazine.com/articles/ai/-neurons-brain-molecular-computer-evolves and
http://www.nature.com/nphys/journal/v6/n5/abs/nphys1636.html)

I have tried to figure out "how each cell of this cellular automaton much be build, so when we put it in a network, the whole exhibits functional behavior." And studying the real neuron seems to be the most logical answer. Also being creative in designing cellular automaton.

The only resource I have about cellular automaton is Stephen Wolfram's 'New Kind Of Science'. He seems very creative in making cellular automaton.

What about you? Do you mind telling me more about your study in artificial consciousness?

Sincerely,
fasisi


tkorrovi
posted 5/30/2010  04:36Send e-mail to userReply with quote
I have reasons to consider that cellular automaton is not unrestricted enough.


fasisi
posted 6/1/2010  05:38Send e-mail to userReply with quote
 
tkorrovi wrote @ 5/30/2010 4:36:00 AM:
I have reasons to consider that cellular automaton is not unrestricted enough.

 
Really? I have read that there is a 'universal automata' that can calculate anything. What is that mean? Do you mind share what you know about this?

Thank you,
fasisi


tkorrovi
posted 6/1/2010  09:04Send e-mail to userReply with quote
This universal means just that it is turing complete. But this doesn't mean that turing machine can emerge there as a result of self-development. Also, we can implement everything in turing machine, but for that again we need some system in turing machine to implement it, thus this alone is not a solution. Thus, turing complete is not yet unrestricted enough.


fasisi
posted 6/1/2010  10:01Send e-mail to userReply with quote
 
tkorrovi wrote @ 6/1/2010 9:04:00 AM:
This universal means just that it is turing complete. But this doesn't mean that turing machine can emerge there as a result of self-development. Also, we can implement everything in turing machine, but for that again we need some system in turing machine to implement it, thus this alone is not a solution. Thus, turing complete is not yet unrestricted enough.

 
What about the IBM's claim that they have build system that resembles a brain of a mouse?


tkorrovi
posted 6/1/2010  20:53Send e-mail to userReply with quote
What claim? It is difficult to search like this. There have been many claims, but the information they give is more than vague. Do you mean this http://news.bbc.co.uk/2/hi/technology/6600965.stm ? "The researchers said they had seen "biologically consistent dynamical properties"", well, this doesn't say much. Also this one is old, and i found nothing about mice brain simulated by ibm, in sciencedaily. I found several similar articles though, but i don't know what of these is about the claim which you talk about.

They don't even know how neuron works, so how can they possibly make a working mouse brain?


fasisi
posted 6/2/2010  07:29Send e-mail to userReply with quote
 
tkorrovi wrote @ 6/1/2010 8:53:00 PM:
What claim? It is difficult to search like this. There have been many claims, but the information they give is more than vague. Do you mean this http://news.bbc.co.uk/2/hi/technology/6600965.stm ? "The researchers said they had seen "biologically consistent dynamical properties"", well, this doesn't say much. Also this one is old, and i found nothing about mice brain simulated by ibm, in sciencedaily. I found several similar articles though, but i don't know what of these is about the claim which you talk about.

They don't even know how neuron works, so how can they possibly make a working mouse brain?

 
Hello,

I am sorry to make you search the article. I should have put more information for you. I am sorry.

And yes, that (http://news.bbc.co.uk/2/hi/technology/6600965.stm) is the article I was talking about.

That article amazed me. Because I know that analyzing a small part of a brain may take many many years. If the brain is so hard to study, why someone can say something like that?

(I've sent an email to one of the researchers and he was reluctant to answer my questions.)

I'd like to talk more about you. If you don't mind.
So, how did you started studying artificial consciousness?


tkorrovi
posted 6/2/2010  11:00Send e-mail to userReply with quote
Well, there is not much to say about me. I also tried to solve this "big problem", that is, how the human mind works, like so many others. The only advantage which i perhaps had was that i had no good perspectives in my life in any possible way, and so also nothing better to do. This enabled me to repeatedly go through that huge pain which everyone has to suffer who wants to solve some extremely difficult problem. If i had any even half satisfactory solution for my life, then most certainly i did not torture myself with things like that. So i understand very well why scientists, etc, don't do that, simply, they can live. I currently work as a software engineer in a uk company which does these eu cordis projects. Well, my system is also proposed under fet, but these things don't go anywhere, as because such research would not give any financial benefit whatsoever, there is no way to push it through. I think there is not much else to say about me.

Last edited by tkorrovi @ 6/2/2010 11:09:00 AM

fasisi
posted 6/3/2010  06:24Send e-mail to userReply with quote
 
tkorrovi wrote @ 6/2/2010 11:00:00 AM:
Well, there is not much to say about me. I also tried to solve this "big problem", that is, how the human mind works, like so many others. The only advantage which i perhaps had was that i had no good perspectives in my life in any possible way, and so also nothing better to do. This enabled me to repeatedly go through that huge pain which everyone has to suffer who wants to solve some extremely difficult problem. If i had any even half satisfactory solution for my life, then most certainly i did not torture myself with things like that. So i understand very well why scientists, etc, don't do that, simply, they can live. I currently work as a software engineer in a uk company which does these eu cordis projects. Well, my system is also proposed under fet, but these things don't go anywhere, as because such research would not give any financial benefit whatsoever, there is no way to push it through. I think there is not much else to say about me.

 
Help me understand you better. What do you mean by saying "The only advantage which i perhaps had was that i had no good perspectives in my life in any possible way, and so also nothing better to do"?

What do you mean by "... had no good perspectives..."?

And "If i had any even half satisfactory solution for my life, then most certainly i did not torture myself with things like that". Do you mean that you prefer not do "artificial consciousness" thing?

And what is "eu cordis"? Is it "Community Research and Development Information Service"?





tkorrovi
posted 6/3/2010  09:36Send e-mail to userReply with quote
> What do you mean by saying "The only advantage...

Well that basically means, nothing to hope for, no good solutions no matter what or how much i did.

> Do you mean that you prefer not do "artificial consciousness" thing?

No, not at all, i'm very interested in such the most basic things. But the fact that these things are very interesting doesn't mean that doing them makes you anyhow happier. Solving these big problems causes a lot of pain, so much that you would be completely broken and as unhappy as can be. Then you build yourself up, and then you go through that pain again, this is only unpleasant, and no hope for anything better. I hope that i solved the biggest problems and no one has to go through that again, but as i'm not a prominent scientist then people ignore me and prefer to think everything from the beginning by themselves.

> And what is "eu cordis"? Is it "Community Research and Development Information Service"?

Yes.


fasisi
posted 6/3/2010  13:24Send e-mail to userReply with quote
 
tkorrovi wrote @ 6/3/2010 9:36:00 AM:
> What do you mean by saying "The only advantage...

Well that basically means, nothing to hope for, no good solutions no matter what or how much i did.

> Do you mean that you prefer not do "artificial consciousness" thing?

No, not at all, i'm very interested in such the most basic things. But the fact that these things are very interesting doesn't mean that doing them makes you anyhow happier. Solving these big problems causes a lot of pain, so much that you would be completely broken and as unhappy as can be. Then you build yourself up, and then you go through that pain again, this is only unpleasant, and no hope for anything better. I hope that i solved the biggest problems and no one has to go through that again, but as i'm not a prominent scientist then people ignore me and prefer to think everything from the beginning by themselves.

> And what is "eu cordis"? Is it "Community Research and Development Information Service"?

Yes.

 
"But the fact that these things are very interesting doesn't mean that doing them makes you anyhow happier"

Really? I thought it is the other way. I think that doing what interest you can make you happy.

"Solving these big problems causes a lot of pain...".
What kind of pain is this? Physical pain like head ache? Psychological pain like 'hurt of rejection'?

"... no hope for anything better."
What kind of hope? What do you hope?

"I hope that i solved the biggest problems and no one has to go through that again, but as i'm not a prominent scientist then people ignore me and prefer to think everything from the beginning by themselves."

Is that what make you feel "no hope"?


tkorrovi
posted 6/3/2010  21:00Send e-mail to userReply with quote
> Physical pain like head ache?

Yes it is like head ache, but much worse, and lasting for a long time.

> What kind of hope? What do you hope?

Well, simple things. Like now, all trees are blooming and all nature is so fresh and green. I wanted to just look at it and fully enjoy it.


fasisi
posted 6/4/2010  07:55Send e-mail to userReply with quote
 
tkorrovi wrote @ 6/3/2010 9:00:00 PM:
> Physical pain like head ache?

Yes it is like head ache, but much worse, and lasting for a long time.

> What kind of hope? What do you hope?

Well, simple things. Like now, all trees are blooming and all nature is so fresh and green. I wanted to just look at it and fully enjoy it.

 
Since you are interested in artificial consciousness, have you ever succeeded in trying to model consciousness?


tkorrovi
posted 6/4/2010  09:16Send e-mail to userReply with quote
> Since you are interested in artificial consciousness, have you ever succeeded in trying to model consciousness?

Well, the biggest problem in training unrestricted systems is that they have no fixed goal. Thus, some combination of input which happens to be the most stimulating for the system, has to be used. What it is, may only be found by interacting with the system. The second problem is that in such systems the success can be estimated only statistically. I got rresults with quite high z score (link below), but this was rather to show the concept.

 Artificial Consciousness ADS-AC project

fasisi
posted 6/4/2010  10:07Send e-mail to userReply with quote
 
tkorrovi wrote @ 6/4/2010 9:16:00 AM:
> Since you are interested in artificial consciousness, have you ever succeeded in trying to model consciousness?

Well, the biggest problem in training unrestricted systems is that they have no fixed goal. Thus, some combination of input which happens to be the most stimulating for the system, has to be used. What it is, may only be found by interacting with the system. The second problem is that in such systems the success can be estimated only statistically. I got rresults with quite high z score (link below), but this was rather to show the concept.

 
Your ADS is interesting. I am going to download the source codes from SourceForge. Before I learn them, would you mind introduce me with your ADS? What is the basic idea? Things like that, that might help me understand your system easily.

This is getting interesting!


tkorrovi
posted 6/4/2010  13:13Send e-mail to userReply with quote
> Before I learn them, would you mind introduce me with your ADS?

Well in short, it's just a system which was derived to be as unrestricted as possible, which means more unrestricted than cellular automata. Why we need a system which is more unrestricted than cellular automata? It is for example because no one has yet succeeded in training a cellular automaton (i mean not recurrent nn) -- its patterns are kind of fragile and would die after almost any interaction.

Last edited by tkorrovi @ 6/4/2010 1:15:00 PM

Bert
posted 7/5/2010  23:23Send e-mail to userReply with quote
 
tkorrovi wrote @ 6/4/2010 1:13:00 PM:
> Before I learn them, would you mind introduce me with your ADS?

Well in short, it's just a system which was derived to be as unrestricted as possible, which means more unrestricted than cellular automata. Why we need a system which is more unrestricted than cellular automata? It is for example because no one has yet succeeded in training a cellular automaton (i mean not recurrent nn) -- its patterns are kind of fragile and would die after almost any interaction.

 
Without insult...Is this for real?




tkorrovi
posted 7/5/2010  23:57Send e-mail to userReply with quote
 
Bert wrote @ 7/5/2010 11:23:00 PM:
Without insult...Is this for real?



 
Yes, but nothing is too good, you know. More unrestricted, but very difficult to train and control. Everything comes with its downsides, this cannot be avoided. And be sure, these dilemmas would cause many head aches in the future, but this too is inevitable, because they are there.

I talked with Fasisi by email, btw, but he kind of disappeared. It's bad that all these discussions are so scattered, especially these by email. It would be much better when everything would be discussed in a forum. Otherwise i have to explain these things to everyone from the beginning again, which i have done quite many times.

  1  2  
'Send Send email to user    Reply with quote Reply with quote    Edit message Edit message

Forums Home    The Artificial Intelligence Forum    Hal and other child machines    Alan and other chatbots  
Contact Us Terms of Use