How do you measure if the LLM actually answered the user’s question, even if the facts it provided were technically true?

When would you evaluate a model without having a "correct" answer to compare it against? (e.g., checking for tone or politeness).

At what stage of the evaluation pipeline is a human absolutely necessary, and where can they be replaced by an automated "Judge LLM"?

If you are using an LLM to grade another LLM, why is it critical to provide a "multi-point rubric" rather than just asking "Is this answer good?"