Why is it harder to A/B test an LLM prompt than a UI button color? How do you account for the "non-deterministic" nature during the test?

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

Guardrails add an extra check. How do you evaluate if the safety benefit of a guardrail outweighs the 200ms latency penalty it adds?

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?"