What is grounding in AI search?
Grounding is when a model answers using documents retrieved at the time of the question, rather than only from what it absorbed during training.
A grounded answer cites sources because it has just read them. An ungrounded one is drawn from the model's parameters, which reflect the state of the world at training time and cannot be traced to a document.
This distinction matters for measurement more than most people expect. If you ask an engine a question and it answers without searching, you have measured what the model remembered about your category months ago, not what a buyer would see today. Any tool measuring AI visibility should be asserting grounding explicitly on every call and treating a failure to confirm it as an error rather than as an answer.
Related
- CitationA citation is a source an AI assistant links or refers to in support of an answer it has given.
- Retrieval-augmented generation (RAG)Retrieval-augmented generation is the technique of retrieving relevant documents and giving them to a model as context before it answers.
- AI visibilityAI visibility is how often, and how prominently, a brand is named by AI assistants when people ask questions in its category.
Want to know where you actually stand on this? Run a free visibility check or try the free tools.