Limits to scalable evaluation at the frontier: LLM as Judge won’t beat twice the data

Best AI papers explained - Limits to scalable evaluation at the frontier: LLM as Judge won’t beat twice the data