Framework

Retrieval-augmented, source-grounded answering

The question it answers

When a manager uses AI to look up contracts, policies or internal reports, how can they make it answer from those documents and cite sources they can check?

Origin
Lewis et al., 2020
Diagram type
Process by phase
Level of analysis
Individual
Topics
AI for leadersThe MBA journey
Glass sphere with a glowing neural-network core, icon for AI for leaders

Diagram

Retrieval-augmented, source-grounded answering

  1. Four steps
  2. Load your documents
  3. The system retrieves the relevant passages
  4. The model generates an answer with citations to the source passage
  5. The human opens the source passage to check

Pitfalls

Pitfall

Treat claims that document-grounded tools produce no hallucinations with caution; research and common sense say grounding lowers errors but does not remove them.

Sources

References

Listed A–Z by author surname, Harvard style. A source with a DOI links to https://doi.org/…; one without links to its page or says plainly that it has no DOI.

  1. Lewis, P., Perez, E. and Piktus, A. (2020) Retrieval-augmented generation for knowledge-intensive NLP tasks. (arXiv preprint arXiv:2005.11401).doi:10.48550/arXiv.2005.11401Source page