TL;DRGroundedness means tying answers to sources — citations and attribution reduce hallucination.
An answer is grounded when it's supported by provided, citable sources rather than the model's memory. RAG enables this: feed documents and ask the model to answer only from them and cite which passage. You can then verify claims against sources. Groundedness + citations turn an LLM from a confident guesser into an auditable assistant — essential for research, legal, medical, and enterprise use.
Key points
Answers supported by citable sources
RAG + "answer only from these docs"
Cite passages → verifiable claims
Turns guessing into auditable answers
Common mistakes
Letting the model answer from memory when sources exist
No citation → can’t verify
Citing sources that don’t actually support the claim
Try it: Explain how citations make an LLM answer auditable.
Example code
<!doctype html><html><head><meta charset="utf-8"></head>
<body style="background:#06040d;color:#e6e0ff;font-family:monospace;padding:20px"><pre>prompt: "answer ONLY from these docs, cite the passage"
answer: "X is true [doc 2, p3]"
→ verify against the source</pre></body></html>