The existing logic inside the brain of a Maggot must be composed of some simple behaviour functions. In order to provide functionality and inate behaviour these behaviours must be described as a set of simple sequences, selections and iterations(repetitions). These are the three fundamental components of logic and computer code! (Dunn 2003). This indicates a strong correlation between brain logic and computer code. There must be a way of implementing a Maggot's behavioural inteligence with code. Once this is implemented, the A-Life Maggot could pass on its source code to children with small modifications in the code. Eventually some Maggot would be born with new behaviours and more chance of successfully reproducing. Isn't this how complex intelligence evolves in the real world!? We just need to convert a maggot into source code. Nightwalk[Edited by - Nightwalk on November 5, 2004 10:56:59 AM]

hehe. Sounds like fun coding. I've tried to wrap my brain round code concepts like that before to build a perfect evolving system.
Didn't manage it :( Taking into account time difference was very hard.
It's easy to perfrom an action based on current parameters & tweak the requirements & result, but taking time difference & memory into account can throw things.
Some might say that if the system were perfect it would evolve these by itself, but people also say that human intelligence & free thought can be explained in the same manner; & I don't buy it I'm afraid.
Looking @ the current movements in computing & technology, ai development hasn't seemed to have moved @ all. Sure there may be some great experiments out there, but if people are starting from step zero every time I don't see this as progress.
I am doubtfull just how complex a selflearning ai system like this can actually become. I certainly don't believe that such a perfect system could ever be created :/
But code is all about creation & there is no seperation between illusion & reality anyway. I think in the long run of ai development some very smart looking ai's may be built. But as far as the ability to rewrite itself on such massive scales goes.. I don't think so :(
Doesn't mean we should stop trying though. Just prompts looking @ it from another direction I think. & that direction would be a perspective we are yet to achieve.
But it certainly is fun to think about such things. & heck in the end what I like seeing is stuff that appears & reacts in an original & cool way. I don't so much care about the inner workings: Single purpose ai does have it's place & I don't think a true multipurpose-uber-ai is possible anyway given current outlooks.
Prove me wrong (or push the boundries in a direction I do not yet understand). I challenge you!

Tallo Ho!

A maggot conceivably came from raw chemicals, back n-billion years ago. We're all made of star-dust after all.

IMHO the problem is not in replicating the maggot-- it's replicating the world in which it lives.

Will

Quote:

Original post by RPGeezus A maggot conceivably came from raw chemicals, back n-billion years ago. We're all made of star-dust after all. IMHO the problem is not in replicating the maggot-- it's replicating the world in which it lives. Will

Barely conceivably. Different people throw about different numbers depending on what they believe (fully random evolution from small molecules to life to humans, some similar process guided by a higher power, or a very literal Creation) but the probabilities seems very small. Of course in some ways we should have an advantage and be able to speed up the process, you'd think.

Quote:

In order to provide functionality and inate behaviour these behaviours must be described as a set of simple sequences, selections and iterations(repetitions). These are the three fundamental components of logic and computer code!

MUST? I'm not sure I'd go that far. "can probably" would be good :)

Quote:

Of course in some ways we should have an advantage and be able to speed up the process, you'd think

You mean we get to use a very fast single processor instead of millions upon millions of them working in parallel? [wink]

I'd be inclined to aim lower than the extraordinary pinacle of evolution which is the maggot. Bacteria, for example. And they need a reason to evolve into a thinking thing - it isn't the nature of evolution to become more complex, only to diverge. That one of the many paths it follows happens to be more intelligent is extremely likely, but your simulation won't know which of the billions of paths it begins with needs to be continued.

It would still, however, be fun to do :)

Quote:

Original post by RPGeezus IMHO the problem is not in replicating the maggot-- it's replicating the world in which it lives.

I am of the same opinion. I am not a professional in the field of ai but I feel fairly certain that it's a Law or at least a Theorem that learning is limited by the environment of the learner(and if it's not, can I get it named after myself).

I believe any algorithm to learn would produce any level of intelligence if given suitable classes(as in school, not OOP).

Quote:

Maggot Intelligence

Isn't that an oxymoron?
[grin]

Quote:

Original post by d000hg Barely conceivably. Different people throw about different numbers depending on what they believe (fully random evolution from small molecules to life to humans, some similar process guided by a higher power, or a very literal Creation) but the probabilities seems very small. Of course in some ways we should have an advantage and be able to speed up the process, you'd think.

The probabilities *are* very small, that's why life isn't rife out there in the great expanse of the universe. And we do speed up the process, considerably, with techniques such as genetic algorithms and neural networks. However, we only have limited processing power and physical storage (the human brain has trillions of neurons), and the structure of the brain is something that has taken 100's of millions of years to evolve. Compared to that we've made incredible progress.

Quote: Barely conceivably. Different people throw about different numbers depending on what they believe

I Think that even if all intelligence in the universe was created by a higher power, something must have created the higher power. So at some point, intelligence must have evolved completely by itself from nothing. There is no other way!!

Quote: That one of the many paths it follows happens to be more intelligent is extremely likely, but your simulation won't know which of the billions of paths it begins with needs to be continued.

Yes it will! It doesn't need to know which ones to continue, the paths will automatically take over by themselves and continue because they are better and more efficient than the other paths! like in natural selection.

Quote: we get to use a very fast single processor instead of millions upon millions of them working in parallel?

Thats true even with a dual 999GHz CPU it would be nothing compared to the entire universe which has billions of billions of things happening simultaneously. Think how many 999GHz CPUs could fit inside the universe?! If we wanted to have a CPU as complex and fast as the universe, it would need to be the same size.

Quote: IMHO the problem is not in replicating the maggot-- it's replicating the world in which it lives.

I think that's true! the world is also very important, but not more important than the maggot, they both need to be there or it won't work.
Maybe we could give senses to the a-life so that it can exist in our world. If we got a computer and gave it a camera, a microphone, a touch-screen, a gas detector and a taste detector it would have the same 5 senses that we do (isn't balance our 6th sense?!?)

Alternatively, Maybe if we want true intelligence we should stop trying to create from scratch and start modifying the human Gene Pool and DNA and create little mutant babies with Einstein brains. :)

Quote:

Yes it will! It doesn't need to know which ones to continue, the paths will automatically take over by themselves and continue because they are better and more efficient than the other paths! like in natural selection.

The problem is that you need something to determine which is 'better' and 'effecient'. In the real world this is typically pre-mature death (as in you don't live long enough to make babies). In a simulation nothing will automatically do this for you.

Without some type of environment which can accurately simulate the conditions _favorite_life_form_here_ would live in, you can hardly be simulating the _favorite_life_form_here. It would be like saying the Sims is somehow a detailed simulation of humans.

Quote:

the world is also very important, but not more important than the maggot, they both need to be there or it won't work.

That first sentence reminds me of a C.S. Lewis book. :) Heh.

Quote:

Maybe we could give senses to the a-life so that it can exist in our world. If we got a computer and gave it a camera, a microphone, a touch-screen, a gas detector and a taste detector it would have the same 5 senses that we do (isn't balance our 6th sense?!?)

This wouldn't be such a bad idea but there is still one component missing: a goal. What would define one iteration of the program better than any other?

Consider this: I have a screen saver where fish are rendered beautifully in a simulated fish tank. One could say that the fish are perfect simulations of the real thing. In this context we would be unable to disprove the claim simply because the environment the fish live in is not capable of testing all aspects of fishness. Can this fish die? "Sure", the developer might say, if the environment were capable of creating the right stimulus.

Cheers,
Will

I'm struggling to see how this thread is related to Game AI. Heck, it's barely related to AI at all... either bring it back to a discussion of intelligence (and how to emulate/simulate it) or I'll be forced to close this thread. Philosophical discussions on biology and religion can be fun... but they are not appropriate for a technical forum on Game AI.

Cheers,

Timkin