There are different kinds of prompt improvements - some have more to do with asking a good question, that will always be relevant. Others - like adding "let's t

There are different kinds of prompt improvements - some have more to do with asking a good question, that will always be relevant. Others - like adding "let's think this carefully in a step by step way to make sure that the answer is correct" - aren't specific to your topic of conversation with the LLM but have been demonstrated to improve results in many cases. That's the part that might go away as models improve.