Tomorrow's prompt managers should learn the same things. They will need to know which AI models to use for which tasks. They will need to be able to quickly evaluate new models they have never used before to determine if they are good enough. They will have to know how to divide complex tasks between different suitable models to produce the highest quality work.
Model assessment will be a skill in itself. But there is reason to believe that it will be easier to evaluate models than humans, because the former are easier to test. A model is affordable day or night, is usually cheap and tends to get even more, never gets bored or complains and returns results instantly.