Sumner sees a pattern of increasing intelligence causing agents to be increasingly ethical, and sounds cautiously optimistic that such a trend will continue when AIs become smarter than humans.
My shamelessly self-serving hot-take: this claim relies on modeling intelligence as a on-factor capability which it is not, and in fact, you categorically cannot generalize from humans to LLM-based-AI when it comes to morality.
My shamelessly self-serving hot-take: this claim relies on modeling intelligence as a on-factor capability which it is not, and in fact, you categorically cannot generalize from humans to LLM-based-AI when it comes to morality.