A software with a God like smartness is one that will on its own keep on figuring out how to optimize bigger and bigger problems. People should be trying to figure out such a learning software mathematically, not wasting time seeing how the human brain connects pieces of information. All we need to do is create one that works, not one that works the same way as the human brain. There's no need to ever make the software in such a way that it forgets some of the bits of information it receives; instead it can duplicate some of bits of information it learned, which inculdes ones it received and ones it previously figured out, in such a way that it figures things out really fast because the important memories are duplicated quite a lot to exist in many different parts of the computer enabling fast connecting of pieces of information. It could do all sorts of connecting pieces of information including making connections from so much information that the humans can't keep up with learning it such as noticing patterns in the large set of images it is shown, which would enable it to optimize minimizing the sums of the squares of the difference in age in days between the age of a randomly picked grade 5 kid it guessed and their actual age based on past experience. Having the smart learning software occur in a supercomputer that can't move doesn't pose the danger of human extinction like robots building smarter robots does. We can probably greatly speed up the rapid advancement of technology by having a clear organized collaborative research group growing itself bigger instead of just waiting for individual people to just happen to read the right pieces of information that they can connect together into something useful for the world. If we want the swarm intelligence research to occur really fast, we will not be lazy and do less training of the humans than we would have done if there weren't a computer to take over the thinking having the humans not even understand what the computer is doing and having the humans not know what they want. It's not known whether a method of skyrocketing math research to use for creating the learning software exists, but that doesn't mean it's not the case that for almost all mathematical statements that a probabilistic argument shows that the probability of is zero, it turns out that there exists a mathematical proof that that statement is false. For example, the problem that asks for each positive integer n whether there are an even number of prime numbers less than 2 ↑↑ n is probably not computable within exponential time but that doesn't show why a proof that there are infinitely many positive integers n such that there are an even number of prime numbers less than 2 ↑↑ n can't exist.