Examples
Worked examples
- Is an instance
A research assistant that retrieves PubMed abstracts via vector search and feeds them to GPT-4 to answer a clinical question
Counter-examples
Looks similar, but isn't
- Not an instance
A plain LLM that answers from its training corpus without retrieving any external documents is not RAG
Editorial commentary
RAG reduces but does not eliminate hallucination: the model can still misquote retrieved passages or invent content not in them. For scholarly use, disclosure should include the underlying LLM, the retrieval corpus, the embedding model, and the retrieval parameters (top-k, similarity threshold).
References
- Lewis et al. 2020 ‘Retrieval-Augmented Generation’ NeurIPS
- Gao et al. 2023 ‘Retrieval-Augmented Generation for LLMs: A Survey’ arXiv
Also known as
RAG · Retrieval-augmented LLM
Machine-readable encodings
Use in your systems
xml
<role vocab="credit"
vocab-identifier="https://casrai.org/dictionary/"
vocab-term="Retrieval-augmented generation (RAG)"
vocab-term-identifier="https://casrai.org/dictionary/term/retrieval-augmented-generation" />json
{
"@context": "https://schema.org",
"@type": "DefinedTerm",
"name": "Retrieval-augmented generation (RAG)",
"identifier": "https://casrai.org/dictionary/term/retrieval-augmented-generation",
"description": "An AI architecture in which an LLM is augmented at inference time with documents retrieved from an external corpus (often via vector similarity search), so that the model's outputs are grounded in retrieved evidence rather than relying solely on parametric knowledge.",
"inDefinedTermSet": "https://casrai.org/dictionary/domain/generative-ai-use-and-disclosure/",
"url": "https://casrai.org/dictionary/term/retrieval-augmented-generation",
"sameAs": [
"RAG",
"Retrieval-augmented LLM"
],
"license": "https://creativecommons.org/licenses/by/4.0/"
}
