Loading…
Glossary term
Tying a model's answers to specific, checkable sources of information.
Grounding anchors a model's answer in specific sources, such as retrieved documents or search results, and often cites them. The model is asked to answer from the supplied material rather than from memory alone. It makes answers checkable.
Grounding reduces hallucination and lets users verify claims, which is why trustworthy AI products cite sources.