if we're purely talking about gpt-3 then it's not sufficient to replace a human because it is not coherent to make business decisions on it's own, it will need human supervisors.

the way i see it, these LLM AIs will keep getting better but they'll actually empower human workers, not replace them. we'll be capable of doing more thinking and less acting jobs. but yeah we might get rid of actors and only value thinkers in the upcoming job market. who knows how this will evolve with our herd mind.