← All 25 termsAI proof

LLM-as-judge

AIUsing a language model to grade the outputs of another model on dimensions that are hard to express as a deterministic check (tone, completeness, faithfulness).

Full definition

LLM-as-judge uses one language model to score the outputs of another. It is the only practical way to grade open-ended qualities (faithfulness, tone, completeness, qualitative correctness). The technique is biased. the judge model shares failure modes with the candidate. so it must be paired with at least one deterministic ground-truth check and a panel of at least two judge models for high-stakes scoring. On this site, LLM-as-judge appears in the numerical-faithfulness-eval as one of three scoring lanes alongside exact-match and calculator grounding.

See this term in action.

The /proof page maps every term to the gate or artifact that implements it on this site.