If there really were no magic, we'd have real AI by now.
That's not at all clear. It could well turn out that more processing power than is easily available is necessary to run the experiments that would give us the answers to how intelligence works.
By virtue of P probably not equaling NP, I think the "more processing power" argument is ridiculous. To think up a new math theorem using exhaustive search of the space is simply unthinkable, no matter how much computing power you have. Some currently unknown form of creativity must be applied to reduce the search space considerably.
I think it possible that we will not understand intelligence until we can run and tweak intelligent software. Fortunately, we don't have to do anything like an exhaustive search of the space because we have a working example of intelligence that we can copy without having to understand how it works. Doing this will require very large amounts of computing power, but not computronium.
Fortunately, we don't have to do anything like an exhaustive search of the space because we have a working example of intelligence that we can copy
My point was just that anything besides an exhaustive search of the space is applying some sort of "creativity magic."
Also FWIW, I believe that in order to simulate a brain on a computer we will essentially need to know how it works, which we are no where near. That is my opinion as an AI researcher, but there are certainly others who know more than me and disagree.
That's not at all clear. It could well turn out that more processing power than is easily available is necessary to run the experiments that would give us the answers to how intelligence works.