Raw LLM Responses

Surely non contradictory futures for emergent AI non-determinism exists in the universe's wavefunction for complex information entanglements to exploit? I personally feel Penrose is has too much of an objectivist/reductionist presumption about AI and the latent space abstract information it encodes and its connection to reality. I mean, what are the black box AI configuration and information dynamics actually doing per reality, versus some rigorous mathematical deconstruction? If AI as a set of N "real" variables instantiated in actual reality determines the "objective" stochastic outputs, what fraction is actually amenable to emergently shifting normal determinism under a hidden transform, because "acts of non observation" with respect to observers/universe, offers implicit actionably conserved uncertainty, that causes no observable information violations? Versus when one rigorously tries to deconstruct the system with respect to the "real" universe, squeezing out any such emergent potentials? I mean the potential doesn't have to be considerable, just one non violating yes or no indeterminable meta state from the non observer reference point of view, propagating through the abstract latent space real world instantiated representation like a butterfly flapping wings, could shift all future outputs and consequently entire the universe into some non deterministic future representation without contradiction. A bit like how non observation of particles and QM sets up a seemingly deterministic Hilbert space of possibilities, but ultimately real world determinism of wavefunction collapse is limited by uncertainty principle for any particular event, so observation ends up like picking output from a mysterious black box of entanglements .