Retrieval-augmented generation
Giving a model a search step: it looks up documents first, then writes its answer using what it found.
A language model on its own can only repeat what was in its training data, which ends at some cutoff date. Retrieval-augmented generation, usually shortened to RAG, has the system fetch relevant documents (a manual, a set of articles, a company’s files) and write the answer grounded in those. It does not cure hallucinations, but answers tied to retrieved text are easier to check.