But these are philosophical questions, not scientific questions. "Could God be hidden in black holes" is unknowable in the same way that "Is God a flying spaghetti monster?" is unknowable. It's not an interesting scientific question because it has nothing to do with the scientific problem of how black holes work, but with the philosophical problem of whether there is a God.

And just because engineers don't usually understand what their AI models do 'under the hood' doesn't mean they can't be understood. They are fundamentally just very complex decision trees and you could in principle see why each decision in a model was made in a certain way. It'd just take a very long time.