Even Robots Can Be Heroes (Hamilton's Rule by evolutionary algorithms)

[shernren]

Or, a really elegant experiment involving evolutionary algorithms demonstrates that kin selection can produce altruism. But the article's headline is way cooler.

Even Robots Can Be Heroes - ScienceNOW

And yay, the original article is open access! Way to go PLoS Biology.

http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.1000615

 

[shernren]

They even had pleiotropy and epistasis - all with only 33 genes. Frickkkk. Interestingly Hamilton's Rule still held, even though it was originally formulated with the assumption that pleiotropic and epistatic effects were absent.

(Pleiotropy: one gene affects multiple traits; epistasis: one gene's expression is controlled by other genes.)

One of the keys seems to be that the genes coded for connection strengths in a neural network. Evolutionary complexity depends on both intra- and inter-organismic feedback loops.

 

[Smidlee]

Interesting except animals are not robots: Evolution Of Animal Personalities ( It seems even evolutionist thinks animal including insects are more complex than simple robots)
We know we have a large impact of animals and sometimes guilty of imposing our way of thinking on them but robots and computer programs are 100% products of our own minds. It's too easy to get the results you are looking for.

 

[Mallon]

Interesting except animals are not robots:

It's funny to hear a creationist say this. How many times have you heard a creationist compare the complexity of life to that of, say, an airplane or some other machine?

 

[Smidlee]

It's funny to hear a creationist say this. How many times have you heard a creationist compare the complexity of life to that of, say, an airplane or some other machine?

I see "Life" a lot more complex than man-made machines, the same with robots/computers vs animals/brain. Is it possible for man to create his equal?
p.s I love math and thinks it's very useful yet I still feel life is a lot bigger than just mathematics.

 

[shernren]

Sure, animals aren't robots. Robots make good models, though. They come in really handy with the ethics committee in particular.

 

[Smidlee]

Sure, animals aren't robots. Robots make good models, though. They come in really handy with the ethics committee in particular.

I wonder just how much information has been created by this program. I have a strong feeling it nothing but a "trial and error" search engine which is loaded with information to get the results the programmer wanted. If the program failed they would continue to rewrite the program (by humans not by robots) until they get the results they wanted. The same rewriting a chess program until it beats a chess master. At least a chess program can be put to a real test by playing a living person. Even a good chess program will sacrifice in order to gain a few moves later.
One thing about AI as long you keep thing very simple (limited) like a game of chess with limited amount of space, 16 pieces per side, absolute combat results,etc. the AI can be programmed very strong since it can quickly search a large number of possible moves. But you add in a few variables the AI quickly becomes very weak. There are many examples of TBS games that shows this; for example Civilization.

Evolutionary algorithms is nothing but a fancy name for "trial and error" search engine.

 

[shernren]

Their original research paper is right there in the OP. You can see exactly what they did. You can even email them directly and ask them the very questions you asked me.

 

[shernren]

Actually, this question can be answered directly from the research. The researchers showed that the robots evolved altruism when Hamilton's criterion said that it was advantageous for them to, and that they didn't evolve altruism when it wasn't.

If the robots had simply been preprogrammed to display altruism after a set time, then they would always have done so no matter whether it was advantageous or not.

Had the robots been preprogrammed to display altruism when it was favorable? The problem is that no one copy of the program would have been able to determine whether it was favorable to be altruistic or not. The programmers can set probabilities for various outcomes, but each copy of the program only knows either that it gets to propagate its "genes" or that it doesn't. Each copy of the program therefore cannot know exactly how favorable it is to be altruistic.

Therefore, the development of altruism in the robots was not preprogrammed by the programmers into the robots' code.

 

[Assyrian]

Is it possible for man to create his equal?

Any women want to answer that?

 

[Smidlee]

Actually, this question can be answered directly from the research. The researchers showed that the robots evolved altruism when Hamilton's criterion said that it was advantageous for them to, and that they didn't evolve altruism when it wasn't.

If the robots had simply been preprogrammed to display altruism after a set time, then they would always have done so no matter whether it was advantageous or not.

Had the robots been preprogrammed to display altruism when it was favorable? The problem is that no one copy of the program would have been able to determine whether it was favorable to be altruistic or not. The programmers can set probabilities for various outcomes, but each copy of the program only knows either that it gets to propagate its "genes" or that it doesn't. Each copy of the program therefore cannot know exactly how favorable it is to be altruistic.

Therefore, the development of altruism in the robots was not preprogrammed by the programmers into the robots' code.

Sounds a lot like a chess program. A chess program or any computer program doesn't understand the concept of winning or losing let along athiesm , these have to be included by the programmer. (Chess program will play the same with a losing position as with an advantage one) Again you have to seriously limit the number of variables in these "trial and error" search engines so it can find the answer in a reasonable amount of time.

 

[Brad Watson Miami 1]

shernren,

I like seeing "evolutionary algorithms" on ChristianForums.com. I've discovered Seal #4: S=19 Theory (18.6 algorithm/fractal). I've got "7 seals" revealed as 'beyond Einstein theories' on the Unorthodox subforum - check it out or google these, I'd like to get your comments.