How is the prompt provided to the LLM in a RAG system different from a standard, non-RAG prompt? — AI/ML Interview Question