Posts
Genetic algorithms are strange, almost creepy, but also remarkably cool. This, for instance, is a remarkably cool project by Roger Alsing which seeks to evolve an approximation of the original Mona Lisa using just 50 semi-transparent polygons.
Essentially, to use a genetic algorithm, you only need know what you're looking for in a solution to a problem - not how to solve it. The software does the rest.Genetic algorithms work using a set of parameters to control their output. By running the algorithm with a population of sets of these parameters and selecting those sets that give outputs that a re closest to the desired output, a "natural selection" routine breeds the next generation of parameter sets (adding slight mutations as it goes) and the genetic algorithm runs again. Eventually, a stable design emerges.
I can't help but notice that creationists are very quiet about genetic algorithms. Is there an "elephant in the room" here, I wonder?
Not making a creationist argument here Jon (if there even is such a thing), but wouldn't what this chappy did support some kind of intelligent design theory?(again, I'm not a subscriber, but for the sake of argument!) Surely it demonstrates that random changes can form recognisable or desirable end results when it's being compared to (and subsequently accepted or rejected) a pre-designated ideal?
ReplyDeleteI see your point. I think the keyword here is "random". The algorithm is, for want of a better word, "blind" rather than participating in a conscious design process.
ReplyDelete