G isn't so much the "scientific name for general intelligence" so much as one explanation for the positive manifold of intelligence. There are others: mutualism and sampling.

I'm not sure you're right about the trainability of tests, either.