eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.
RAG — Retrieval-augmented generation — fetching relevant documents first and pasting them into the model's context so it answers from your data instead of memory.
Why it matters
Uber's internal on-call copilot Genie improved answer quality by refining its RAGRetrieval-augmented generation — fetching relevant documents first and pasting them into the model's context so it answers from your data instead of memory.Full definition → pipeline, raising acceptable answers 27% and cutting incorrect advice on security/privacy queries.
articleWhen Knowledge Changes: Metamorphic Testing of RAG Systems with MutationsJinhan Kim, Samuele Pasini, Paolo Tonella
articleUniversal Pathologies, Conditional Consequences: A Triple-Robustness Analysis of RAG for Multi-Hop TraceabilityMeftun Akarsu, Burak Ozdemir