Citations (source attribution)
Links from each claim in an AI answer to the passage it came from, so a reader can verify it.
In a RAG system the model is asked to mark which retrieved passage supports each statement, and the interface shows the document, page or quote. Citations make grounding visible and speed up checking, but they are generated too: a model can cite a real document that does not say what the answer claims. Verifying in code that the quoted text exists in the source closes that gap.
Related terms
- Grounding: Tying a model’s answer to specific supplied sources, such as retrieved documents or search results, instead of its training memory.
- RAG (retrieval-augmented generation): Finding relevant documents first and putting them into the prompt, so the model answers from them.
- Hallucination: A fluent, confident answer that is false or not supported by the sources.
- Faithfulness (groundedness): A RAG metric: the share of claims in an answer that are supported by the retrieved sources, not by the model’s memory or invention.
