As a comment to my last post on the different level of intellegence, i would like to point out that the classification is really just one dimension of human intellegence. It is just looking from behavior point of view. It is not about how we get the intellegence. I touched on the idea that the string theory has brought in, like what if our soul is from a different dimension?

I am also thinking about how to actually demonstrate the behaviors. I mentioned that DARPA race winner Stanley has some learning capability. However, what I would like to do is to demonstrate the behaviors at a lower level, less specific to an application. I am thinking about writing some programs to mimick the capability of a 'body' and then build a structure around it that can 'behave' and take input from the environment. If such a system can be built, then we can actually see how a program can learn.

Eventually, such learning capability has to be built into hardware to get the same performance as today's comptuer program, but I would think we can wait until we see learning program first to worry about different implementations.