Cited but Unused: Parametric Provenance in Retrieval-Augmented Generation
This arXiv paper measures whether citations in retrieval-augmented generation actually reflect retrieved context or instead come from a model's parametric memory. Using counterfactual value substitution, it finds that 1.2% to 15.7% of citable claims across model families derive from parametric memory, with rates rising with model size within a family. Standard natural language inference verification performs no better than chance, even when restricted to the cited passage. Comparing ten attribution methods, the paper finds that the contrast between a model's output distribution with and without context best separates provenance, but the resulting detector remains too inaccurate to act on.
It provides a concrete, cautionary evaluation of citation reliability in RAG systems, showing that standard verification can miss provenance misattribution and that current detectors are not yet actionable.
Comments & Questions (0)
No comments yet
Be the first to comment!