LLM's are trained in much the same way, so while your point stands, most/all of the tips here are going to be useful for LLM's for at least a year or so.

If a tip was like "use XML tags to give clarity to the model," then it wouldn't be sustainable.