this is a tradeoff for being able to answer anything at all

open ai remarks that when it wants the bot to only answer something its 99% confident about say it just never answers because theres always that one part of the answer you arent sure is right.