From Here to Human-Level Artificial General Intelligence in Four (Not All That) Simple Steps – Info AI

Thanks for the interesting read!

I think you are correct that it’s not more power we need, but more theory! Also agree that once a good theory emerges (one that shows some explanatory power and is testable), AIG will arrive very quickly.

I think we can move forward faster if we simplify the goal: understand how the human brain does what it does. If we can do that then further goals like super-intelligence or ability for an AI to explain its decision can follow. (Humans are really poor at explaining our decisions of course, so making that a criterion for “intelligence” is expanding the scope of the problem of understanding how human brains work.)

In fact, we might simplify the initial problem even further: can we understand how an animal’s brain works (like a dog perhaps?) That strips away the symbols/semantics issues (I assume) but still leaves the capabilities for generalized problem-solving and learning, as well as integrated consciousness.

I am not sure that knowing more and more detail about how the neural circuitry of a human or other animal works is going to result in a breakthrough though it might help. As you say, we need a decent theory.