Pulling this up from a subthread: I currently don’t see what the material difference is between this scheme, vs. the following much simpler scheme:
Humane FAIs simulate many possible worlds. (For better coverage, they can use quantum coins to set whatever parameters.)
They find instances of humans about to be killed (by anything, really, but e.g. by UFAIs).
They then extract the humans from the simulation and let them live in the world (perhaps with a different resource cap).
Pulling this up from a subthread: I currently don’t see what the material difference is between this scheme, vs. the following much simpler scheme:
Humane FAIs simulate many possible worlds. (For better coverage, they can use quantum coins to set whatever parameters.)
They find instances of humans about to be killed (by anything, really, but e.g. by UFAIs).
They then extract the humans from the simulation and let them live in the world (perhaps with a different resource cap).