Comment by yunohn

This entire thread is questioning why OpenAI themselves use repetitive negatives for various behaviors like “not outputting JSON”.

There is no magic prompting sauce and affirmative prompting is not a panacea.