Large language models (LLMs) can give us inaccurate or out of date answers. Suppose we can provide additional contexts with our query, then it will be much easier for a LLM to answer questions — because we’ve already included the answer in the context it only needs to extract and summarise the context text.