Artificial intelligence

[infamous]

In summary, AI is a thinking machine. That's all. We can develop a computer that functions similarly to the human brain but, with its machanical limitations, it cannot "live."

 

Exactly Linda; These machines may think, but untill they can participate on all levels of activity, they can not be thought of as alive. We should examine what it means to live, defineing life in terms of active participation is, I believe central to the to definition.

[CarGrrl]

As for intelligence it's a dicey term itself, just what is intelligence? On the computer side there's self generating code. Give a computer a task, and let it work out its own code. Almost 40 years ago a researcher at MIT gave a computer long list of integers and just let the computer figure out what to do. After a few days it figured out how to add and subtract, then a couple days later it figured out multiplication and division. Then it got all weird. It started working out new operations faster and faster eventually coming up with equations one after the other. What's more these equations were so far out it took a few mathematicians a good while to figure out if they were real. In a few cases what the computer came up with was some very obscure stuff, but entirely valid. Odd thing was it was all in the field of number theory. No calculus, geometry, etc

 

So a computer can come up with some pretty nifty stuff on its own in the field of math. Then there's alphabetizing a long list of words. Back in the 80's another MIT researcher gave a computer a big list of words and the alphabet and let it figure out how to alphabetize. What the computer came up with was significantly faster than any human construct algorithm. Also it made no sense to anyone at first. Was seemingly a bunch of jibberish. Yet it worked.

 

Of course figuring out stuff isn't exactly much, there's the matter of communication. And currently computers are not very good at understanding language. While diagramming a single sentence is one thing, working out what's what in a series of sentences is another. And telling someone what War and Peace is about? Yet another. Not that there's any lack of people working on it.

 

If a computer can understand human communication, it should at least in theory, be able to mimic human communication. And if you can carry on a conversation with a box, as long as the conversation coming from the box isn't from preselected possible output, then what's the difference between the box and a person? The box might not dream but it can at the least say how it's doing and what it's doing.

 

Personally I see the big problem being the architecture of computers itself. All binary and the same old stuff it's been from the start. If someone came along with a computer that used base 3 or 4 things might get hopping in a big way. Better bet I think is quantum computing. Which would allow for operations that standard computers can't do. Nevermind the speed issues. Also there's some theoretical possibilities that are interesting. Like imagine a computer that can rearrange it's own CPU or alter its instruction set. Yes on paper a quantum computer is years away, but then one person with the right idea may be sitting in a lab right now and have a full scale quantum computer built by the end of this year.

 

As far as I'm concerned it isn't a matter if we can build HAL but a matter of when. According to 2001 HAL should already exist. That HAL doesn't exist is as much the fault of too much hardheadedness on research projects as it is Arthur C. Clarke's being overly optimistic. But who knows perhaps by 2010 there will be a HAL-like computer. As for wether or not it will be a true intelligence, if it carries on conversation about whatever topic is brought up then what else could it be called? Hard subject is, would it be sentient? That's tough to say. I suspect that isn't something at all programmable, but rather an arrived at state. As for how it'll be determined, I get a feeling it won't be a problem. Though I don't necessarily believe that will give any insight at all into human sentience or mental processes or some such. After all, it'll still be a computer.

[Chris The Great]

One of the huge problems in creating an AI seems to be that computers can manipulate symbols such as numbers extremely easily and quickly, however recognizing patterns, such as faces and voices which humans do instantly and subconciously is extremely difficult for our current programming languages.

 

I think that our current hardware can and will work for artificial intelligence. I believe that the programming languages themselves, as well as the computing speed of even the fastest supercomputers being completely inadequite, are what is limiting the developement of AI. It will probably take a big breakthrough to create a new language that will be able to create AI.

Also, currently work is being done on making binary more than 1's and 0's, and include different degrees such as 0.1's and such. This is supposed to be possible with our current hardware technology, and would definatly increase speed and probably make an AI much more efficient than a simple on-off binary system.

 

The other option is that computer power will increase so much in the next few decades it doesn't matter what programming code we use, we could create a simulation of a brain and run that, and it would become a concious being. I think something like that would require vast amounts more computing power than an actual intelligent program however. Perhaps this will be an intermediate step, and the AI will write it's own programming language and source code.

 

Of course, once the AI is able to do that, it can rapidly improve it's programming and efficiency. It could then redesign it's hardware with some nifty new nano-tech it onvented, and become even faster. It could then improve itself more, and so on and so forth. Which would be really cool, until it starts to 'kill all humans'.

 

This is definatly an interesting subject, if I wasn't doing a presentation tommorrow on the possible consequences and whatnot of AI I'd say more. But I still have lots of planning to do, because I tend to leave actually work to the very end. Sometimes homework is fun....but then you realize you have to do it and suddenly it's a chore.

[lindagarrette]

Exactly Linda; These machines may think, but untill they can participate on all levels of activity, they can not be thought of as alive. We should examine what it means to live, defineing life in terms of active participation is, I believe central to the to definition.

Being alive as animals, anyway) means having the survival instinct, which means feeling pain, satisfaction, panic, feeling of loss, lust, and other effects of neurotransmitters. Artificial intelligence is certainly capable even now of much faster and complex algorighms than the human brain. But any technology that would create a mechanical life form is way beyond forseeable at this time. I don't think any one has even speculated on the possibility other than to suggest it.

[infamous]

Being alive as animals, anyway) means having the survival instinct, which means feeling pain, satisfaction, panic, feeling of loss, lust, and other effects of neurotransmitters. Artificial intelligence is certainly capable even now of much faster and complex algorighms than the human brain. But any technology that would create a mechanical life form is way beyond forseeable at this time. I don't think any one has even speculated on the possibility other than to suggest it.

 

There may be one other way to approach this difficult impediment. With the virtual reality systems, which by the way are becoming increasingly more life like all the time, it may be possible to simulate a life like enviornment for AI to evolve within. This is, I will admit a streach, but a possible answer for creating the format for a developing AI evolutionary enviornment. An intellegent machine bathed in virtual stimuli could possibly develope a very life like world view of it's existence. Just a thought Linda.

[whoa182]

Machine learns games 'like a human'

 

12:17 24 January 2005

NewScientist.com news service

 

 

http://www.newscientist.com/article.ns?id=dn6914

 

check that out:D

[infamous]

Machine learns games 'like a human'

 

12:17 24 January 2005

NewScientist.com news service

 

 

http://www.newscientist.com/article.ns?id=dn6914

 

check that out:D

 

Very thought provoking, inductive logic programming. Learning by watching, imitation, sounds a bit like childs play. This is interestingly, exactly how young humans learn about their world. We could be on to something here!!

[lindagarrette]

Very thought provoking, inductive logic programming. Learning by watching, imitation, sounds a bit like childs play. This is interestingly, exactly how young humans learn about their world. We could be on to something here!!

This is an example of neural networks' boolean logic with learning thresholds. It's a good definition of many AI advances. The biological brain works similarly or maybe just like.

[pgrmdave]

Are neural networks software or hardware? I think I read somewhere that they were a little of both, but it was a long time ago.

[lindagarrette]

Are neural networks software or hardware? I think I read somewhere that they were a little of both, but it was a long time ago.

Neural network algorighms are processed on computer platforms. The platform is not necessarily specific to the neural net.

[Buffy]

Neural network algorighms are processed on computer platforms. The platform is not necessarily specific to the neural net.

The hardware and wetware platforms are significantly faster than the von Neumann/Turing Machine based systems though....

 

Cheers,

Buffy

[Turtle]

I propose the "Turtle Test" for artificial intelligence. Proposition: A machine may be accorded the quality of intelligence if it spontaneoulsely produces a poem similar to that which follows, & then takes steps to bring its poem to the attention of others. :hyper:

 

Morgan's Poem

by Turtle George

 

Not long ago and just nearby,

I chanced upon a tiny fly,

Who sat upon a tiny mat,

Upon his head a tiny hat.

 

How very curious then thought I,

To find a hat upon a fly,

And that fly upon a mat,

I'd seen no stranger thing than that

.

I bent right down and spoke to it,

And found it minded not one bit,

It spoke right back and greeted me,

And asked that I should grant its plea.

 

"You see", it said, "I've lost my book,

It's not in any place I look."

And so it was I searched around,

Up in the leaves and on the ground.

 

The book I found hung on a twig,

As you might guess it was not big,

I picked it up quite carefully,

And looked up close the best to see.

 

I wished to know what flys might read,

So with my eye I drew a bead,

And there upon the tiny page,

A list of poems for any age.

 

I gave the fly the book right then,

Amazed at what I'd found within,

Some new learned things I can't deny,

Because I chanced upon a fly.

[alexander]

Nice work Turtle, I'll talk to Tormod if you dont mind, and maybe if you are ok with it, we will put this poem on the next newsletter. If you dont mind. Talking about mind, i think human mind works wonders, whether it is from errors within our system or its true and undoubtable sophistication obstraction is very much possible and is commonly displayed in a form of art, whether it is painting, music, poetry, graphics, prose or anything in between, anything that requires abstract thinking (or commonly used thinkig outside the box) it amazes me, true AI must include that humanly function, and no matter how sophisticated the AI is right now, it is far, far away from working like a real mind...

[TeleMad]

... i think human mind works wonders, whether it is from errors within our system or its true and undoubtable sophistication obstraction is very much possible and is commonly displayed in a form of art, whether it is painting, music, poetry, graphics, prose or anything in between, anything that requires abstract thinking (or commonly used thinkig outside the box) it amazes me, true AI must include that humanly function, and no matter how sophisticated the AI is right now, it is far, far away from working like a real mind...

 

Don't want to sound too much like FreeThinker, but aren't you anthropomorphizing intelligence a bit there? You seem to be defining intelligence as human intelligence: something has to be able to do all that the human mind can do before that thing can be considered intelligent. Might be somewhat like claiming dogs can't run because they don't run the way humans do.

[Turtle]

Yes you may use my poem. Yes we are anthropomorphasizing/anthropomorphizing...uhhh, making human. We are human so we use human standards. You make it sound as if it's a mistake; not only is it no mistake, it is an imperative. :hyper:

[TeleMad]

Yes you may use my poem. Yes we are anthropomorphasizing/anthropomorphizing...uhhh, making human. We are human so we use human standards. You make it sound as if it's a mistake; not only is it no mistake, it is an imperative. :hyper:

 

So chimpanzees and dolphins don't possess intelligence?

 

And are you asserting that the acronym AI stands for Artificial Human Intelligence?

[Turtle]

I didn't mean to, but yes...I so assert. I don't deny the animals' intelligence, but it's not our intelligence. Since it is we humans here conspiring to design machine intelligence, we can'tinclude any other animals' intelligence untill they post here as well. (insert Flipperesc squeaking here) :hyper: