Neophyte, in an artificial world it could be better if also some of your behaviour patterns were inherited, you could think of this inheritance as learning in childhood (we don't need to and we can't simulate everything correctly).

Visit our homepage: www.rarebyte.de.st

GA

Edited by - ga on May 18, 2000 7:12:00 PM

Isn''t instinct inherited?
And where do you draw the line between instinct and knowledge?

The_Minister
1C3-D3M0N Interactive

Yes, instinct is the part of behaviour which is inherited.

Visit our homepage: www.rarebyte.de.st

GA

quote:

---

Original post by Mark_T
Just a question-I''m not sure where this is going.
Minister, what do you mean by ''matrix''?
I''m guessing you mean the layout of a Finite-State-Machine(they do look like matrices, don''t they?)

---

FSM''s do not look like matrices. A matrix is a two dimensional array which can have certain arithmetic operations performed on it. An FSM is a way of modelling something by saying it has a finite number of states, and certain rules for traversing between them. It can be represented by a diagram which has circles with the names of states inside them and lines joining the states to indicate transition rules.

------------------------------
#pragma twice

Dire , sorry to be so negative, but I have to question the point of "emergent intelligence" for your ai opponents when it sounds like you know exactly what you want to "emerge". I mean, why wait around and hope that revenge & evasion desires evolve, when you could just program them in directly?

Are you hoping that they will evolve other, possibly unforeseen desires?

Edited by - Eric on June 23, 2000 12:16:46 PM

I lost my original scathing post about impossibility and the such. The datasource went down! could this be a sign? Anywhos, my original post tried to condemn the idea that real ai is possible due to its inability to model every aspect of existence and predict the impact of evolutionary behaviors, but then it occurred to me that it''s the exact way of life for us, why can''t it be for machines? Not being able to predict what will insure 100% survival is a good thing. In fact, you lose sight of the connection between action and survival, it''s impossible to predict that without disproving heisenberg''s crazy quantum formula. thousands and millions of evolutionary behaviors later, we don''t know what is good for survival anymore. But then again, tracking discrete states and changes in them is the kind of stuff digital computers are good at. So that might be a way to model and predict human evolution. But as far as artificial life, just pick something that looks like it won''t hurt, and cross your fingers! Now, what about loss of desire? how would you introduce jadedness to a machine, without understanding what it means in the human counterpart? What about teenage angst? not all of us go through it when we are young, but some do. How would you model that? I guess digitizing an analogue world is possible, but what kind of computing power does it take?

quote:

---

Are we talking about trying to create a quine ? Which (as I learned in another post)is a program that writes it''s own code . Accept maybe have it change and compile itself .

---

one of our assignment in a language class was to write a computer program that prints out every line of code as it is currently executing in memory. The best anybody did was think about it. how wild is that! I''m definitely looking forward to taking that prof''s compiler class.

okay, i''ve lost track.. quits

cheers,
muffin

Hi (My very first post)

First of all i thing an entity dont need any instinct in the beginning becasuse it will be compared with a robot or an automaton.
I agree 100% with Electron. I think that we only need initial random matrices and let''s the nature do the job.....
what is this job? well this job is mutation, evolution, reproduction(mix matrices of the parents), survive of the fittest and some day that instinct will born in a new generation and life will go on.
We are not modeling human life because is very complicated(emotions, feeling, etc, etc, etc millions of variables and a genome very long and complicated) but we can model a BACTERIA o something like that with a minimun functions(eat and reproduntion or something like that).
That can be the base of a proyect like that electron is thinking.

I''m very interested in Electron proyect because i want to simulate the same thing. If you(Electron) think I can help email me(ahumaro@yahoo.com).

Last of all, I will put this link i found last week: http://www.geatbx.com/docu/algindex.html in that, is explained an algorithm of Artificial Life: An initial generation of matrices, then a cile of evolution, mutation, reproduction and a comparition of fit...... I hope you find that stuff useful.
I think is a very complicated article but is very interesting.

Ahumaro

Hello Everyone.
Let me just say that I just saw this post for the first time and just read through most of it (more like half, took me about half an hour)

AI has always intrigued me, and this term I just finished an intro class to neurology and a Brain and Behvior class (all about evolution and behavior)

First of all, in my B&B class we were shown a clip from a program that an evolutionary theorist programmed. I was quite impressed when I saw it. I think the guy''s name is Simms or something (I remmeber it being ironic), and he created a 3d virtual world with food. The basic plot was the creature had to find food and eat it to survive, and it would reproduce with slight mutations as in the real world. Eventually (it didnt take too long cause 1 year in the rela world of evolution could equal 1 second in a sim world if u want it to) the creatures formed bodies that were most capable of finding the food. Many of them developed flagella like figures so they can swim to get food. It has been done and I have seen a video of it.

I think the important element to remember that I havent seen mentioned (I might have missed it) is that evolution has no goal. You can''t program an arm or a tail or anything at all because those are just results of mutations in our world. We were never meant to look like we do, we just happened to end up like this, not that im saying this is the end. I would consider us at the top of any evolutionary "ladder". I think what needs to be done is:
A) MUST have a physical world established, real or simmed
B) Some sort of heredity with slight mutations
C) Competition and/or challenges

and I think we must just program the beginning of life, very very very simple. The first life form was single cellular. In theory if we can code this cell with the ability to reproduce and mutate and interact with the physical environment we set up, it should evolve on it''s own to better suit itself for the environment it ahs been put in.

And one more thing, aobut the AL taking over your comp and the whole world or whatever, that''s not going to happen because what we''re taling about here is a simm. The only thing the AL will be "concious" of is the world it''s in, the program itself and nothing out of it. It wont know we coded it, it''ll probably be wondering where it came from as we do, assuming any of this is possible at all.

Just thought I''d share my thoughts on the subject, I find it very interesting.

Hey, just registered my name.

I''m the one who wrote that last post if anyone was wondering

Creatures SHOULD inherit knowledge from their parent(s). Its just like they were taught.
-M

Saráth evenü, cán el-sayü liné!