Thought Experiment

(Random aside: this is post #150...go me.)

For a while now I've been intending to explore the world of AIs through computer programming. I wanted to write some script that simulated people would follow and, assuming I wrote it right, prosper in their simulated environment. Lately, I've been interested in finding some very "general" way to do this, instead of writing a script that applies very specifically to the world I'd created.

I hadn't really gotten around to fleshing out this idea too much, until I found myself bored to death in philosophy today. My brilliant (like I said, "go me!"), if staggeringly complex idea, was to right a script that would right the script for people. This means that the programmer (me) wouldn't know anything about the world itself, only the very basic mechanics of it. They (I) would, in essence, program a programmer that would then learn about the world from the automatons in it and have to direct them from this information.

Thinking about it, I realized this went beyond Fun With Algorithms, it was about learning itself. Sweet. However, my first attempts at coming up with a framework for this programmed programmer made me realize it will have to be horrendously complex. If only I had tons if time (i.e. no homework) so I could do this...