Glossary term
Grounding
Constraining a model’s output to information in supplied sources, so that claims can be traced and verified.
A grounded answer is one whose statements are supported by retrieved documents, data or tool results, ideally with citations.
Grounding reduces but does not eliminate unsupported statements, so it should be measured with faithfulness evaluation.
Related terms
- HallucinationOutput from a generative model that is fluent and plausible but not supported by its inputs or by fact.
- Retrieval-augmented generation (RAG)A pattern in which relevant passages are retrieved from a body of content and given to a language model to ground its answer.
- Evaluation setA curated collection of realistic inputs with definitions of correct output, used to measure AI system quality.
Turning concepts into a working system?
We help teams choose the right pattern and build it with evaluation from day one.