Source profile · Academic preprint

Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks

This paper describes retrieval-augmented generation, combining a retriever with a generator for knowledge-intensive tasks. It provides foundational context for source-grounded model answers.

Patrick Lewis et al.Verified 15 November 2026Primary source linked
Reader summary

What this source is useful for

This paper describes retrieval-augmented generation, combining a retriever with a generator for knowledge-intensive tasks. It provides foundational context for source-grounded model answers.

  • Clear technical framing of retrieval plus generation.
  • Useful background for source-grounded AI answers.
  • Citable methodological reference.
Limits

What it does not establish

  • It does not evaluate modern consumer AI search products.
  • The experimental systems predate current product interfaces.
  • Retrieval architecture alone does not explain citation selection.
Evidence note

Why this record exists

Use this source for the general retrieval-generation concept, not as direct evidence about a current AI search citation policy.

Open the original source