Walk me through naive RAG and tell me what breaks.

Naive RAG operates in two primary stages: retrieval of relevant documents and generation of an answer based on those documents. Failure in this system generally falls into three classes. First, the retrieval stage may return irrelevant passages that confuse the model. Second, the generator may fail to perform necessary reasoning steps even when the correct evidence is present in the context. Third, the model may suffer from redundancy, such as repeating the same fact multiple times because it appeared in multiple retrieved chunks. Simply stating that the model ‘hallucinates’ is insufficient as it collapses these distinct failure modes into one.