Reposted by Zac A
What you're getting from an LLM is a statistical approximation of what the answer would be to a statistical approximation of the question.
This is why prompt engineering is fake. The prompt (and any input data) is just a suggestion - and helps fool you into thinking that your question was answered.
