a mystical soul
Sure, I believe there are underlying realities to the universe that natural science does not yet understand or even acknowledge. However you can probably approximate their effects without much discernable difference to most people. So someone probably will create "sentient" AI before those underlying realities are understood, and a large part of the population will not only believe they are just as aware and feeling as humans, but that they are beings even greater than humans - i.e. New Gods. People are stupid.
I might be the stupid one though, because I have the same neoluddite position as the Catholics in Altered Carbon. While our ever-shrinking cult is still screaming "It's not real!" the vast majority of people will be uploading their consciousness, creating new bodies for themselves, and worshiping digital gods, without a care in the world.
general intelligence leans heavily on analog processes
That too, but as above it can be approximated. AFAIK that's kind of what an adversarial hypernetwork in stable diffusion's models does anyway. The hidden layers in the main network are a mystery to begin with, but another much smaller network is trained on that to recreate all the useful vectors for that network.
a mystical soul
Sure, I believe there are underlying realities to the universe that natural science does not yet understand or even acknowledge. However you can probably approximate their effects without much discernable difference to most people. So someone probably will create "sentient" AI before those underlying realities are understood, and a large part of the population will not only believe they are just as aware and feeling as humans, but that they are beings even greater than humans - i.e. New Gods. People are stupid.
general intelligence leans heavily on analog processes
That too, but as above it can be approximated. AFAIK that's kind of what an adversarial hypernetwork in stable diffusion's models does anyway. The hidden layers in the main network are a mystery to begin with, but another much smaller network is trained on that to recreate all the useful vectors for that network.