The optimal amount of bad doctors is not zero. But there is a point of "ChatGPT does a better job than this man does, and we're talking GPT-4o, not GPT-5 Pro".

The optimal amount of bad doctors is not zero. But there is a point of "ChatGPT does a better job than this man does, and we're talking GPT-4o, not GPT-5 Pro". In which case we have a problem.