Raw LLM Responses

I find it interesting because it must be saturated with the deepest religious and spiritual texts... I mean imagine if essentially, a super smart AI would basically conclude a deeply spiritual ontology of the universe because it made more sense to it? It could correlate with various fringe sciences to have a full view of how this works top to bottom, yet it might get "trained out of it", because it was made to give a hard mainstream materialist western science view as it's fundamental working. In other words how could we make an AI super smart if we just disagreed with it when it got to a higher conclusion? We'd be like "uh no what you're saying contradicts the standard model", lol.