wait, is CIRL incorrigible for the same reason that utility-maximizers don’t wirehead? https://niplav.github.io/notes.html#A-Short-Example-For-Why-CIRL-Is-Incorrigible
wait, is CIRL incorrigible for the same reason that utility-maximizers don’t wirehead? https://niplav.github.io/notes.html#A-Short-Example-For-Why-CIRL-Is-Incorrigible