ai evaluating ai

LLM-as-Judge: Using AI to Evaluate AI Output

LLM as judge means using one AI model to grade another's output. What the MT-Bench research actually found, where it breaks, and how to prompt a judge you can trust.