Hmm, if I am to take a guess where you’ll be forced to go, it’s that your assumption about the observability of π(h) won’t hold since π(h) will be non-deterministic in a way that prevents us computationally from fully “observing” it, i.e. knowing when it halts.
Hmm, if I am to take a guess where you’ll be forced to go, it’s that your assumption about the observability of π(h) won’t hold since π(h) will be non-deterministic in a way that prevents us computationally from fully “observing” it, i.e. knowing when it halts.
Nope, it’s not that at all. Assuming π(h) is known (at least to the AI) makes the problem easier, not harder.