How engines work
Hallucination
A hallucination is a confident statement from a model that is not supported by its sources — a wrong price, a feature attributed to the wrong company, an outdated fact stated as current.
In more detail
For brands, most hallucinations trace back to something real but stale: a figure that was correct when the source was written. Refreshing the fact where engines can find it is the practical remedy.
Grounded answers with citations hallucinate less, because the claims are anchored to retrievable documents.
How Allswap measures this
Guides that cover this
Related terms
Grounding
Grounding is tying a model's answer to specific retrieved sources rather than to its own recollection. A grounded answer can be traced back to documents, which is what makes citations possible and answers more reliable.
Training data
Training data is the body of text a model learned from. It sets the model's default understanding of a category and which brands it associates with what, before any live retrieval happens.
Knowledge cutoff
A knowledge cutoff is the date after which a model's training data ends. Anything that changed after that date reaches an answer only if the engine retrieves it live at the time of asking.
Ready to Dominate AI Search?
Get started now
Allswap