liveBook · Manning

Fidelity

As explained in The Art of AI Product Development

Groundedness is the degree to which generated content is supported by the information retrieved or provided as evidence.

Groundedness checks whether generated content is supported by retrieved documents. In retrieval-augmented generation (RAG), unsupported speculation or hallucination signals that the model should rely more tightly on its provided sources. section 7.3.2 section 7.4

Evaluating a generated report

For a report on AI trends in SaaS, assess whether its claims are based on retrieved documents. Separately, a short blog-style response to a request for detailed analysis may be grounded but still fail to answer the query adequately. section 7.3.2

How The Art of AI Product Development covers it

How deep it goes
Defines groundedness and applies it as an evaluation criterion, without specifying a scoring scale.
Examples
Tests a report on AI trends in SaaS and contrasts grounding with relevance to a request for detailed analysis.
What you can do after
Evaluate RAG responses for source support using LLM-as-a-judge, while checking evaluation criteria closely.
How it connects
Links groundedness to retrieval-augmented generation and distinguishes it from answer relevance.

Also explained in

Compare how 4 books explain fidelity →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of The Art of AI Product Development. The section links point to the parts of the book it draws on.