>>10026457If you re-model the human brain you will run into the problem that the human brains largely runs on intuition, and not clear knowledge. So to re-model that, you will have to code in "cores of knowledge" that humans obviously have from birth (or else they would be completely disfunctional as babies) and with that you would have limited the AI in what it can accomplish, beause like humans, it would be limited by its intuition.
The problem rather is that, if we accidentally were to create a superintelligence, we probably wouldn't recognize it. It might do things that seem illogical or stupid to us, but really aren't. Like a chimp wouldn't recognize that a human is way smarter than him, we wouldn't recognize a Superintelligence if we saw it. Hell, we might have created one, deemed it a failure, and deleted it again.