Which statement correctly describes Retrieval Augmented Generation (RAG)?

RAG enhances LLM capabilities by allowing them to retrieve relevant information from external, authoritative knowledge bases before generating a response. This significantly improves the accuracy, relevance, and factual grounding of the output, addressing issues like hallucination, without the costly and time-consuming process of retraining the entire LLM. The first incorrect option describes fine-tuning or pre-training, not RAG. The third option is incorrect because RAG expands the knowledge base, it doesn't restrict it, and its primary goal isn't just speed. The fourth option describes a specific application of LLMs, not the RAG architecture itself.