Module 01 Knowledge Check
5 questions. Pass mark 4 out of 5. Answer every question before checking the answer key below, then retry after reading the feedback.
1. What does RAG actually change about a model's behaviour?
- A. It makes the model truthful
- B. It supplies selected source text at answer time, which the model may or may not use correctly
- C. It retrains the model on your documents
- D. It removes hallucination
2. A plain chat model answers a question about your internal policy correctly. What can you conclude?
- A. It has read your policy
- B. Nothing reliable - it may be reproducing a common pattern that happens to match
- C. Your policy is public
- D. It is using RAG
3. Which task is RAG genuinely well suited to?
- A. Exact arithmetic over a ledger
- B. Answering from a defined, changing document set with citations
- C. Guaranteeing legal correctness
- D. Replacing access control
4. Fine-tuning versus RAG: when does RAG fit better?
- A. When the knowledge changes frequently and must be cited
- B. When you want a different writing style
- C. When you need lower latency always
- D. When you have no documents
5. The most common early mistake in RAG projects is…
- A. Choosing the wrong vector database
- B. Assuming retrieval quality is fine and blaming the model for bad answers
- C. Using too few embeddings
- D. Writing long prompts
---
Answer Key and Explanations
Check these only after attempting every question.
1. B - It supplies selected source text at answer time, which the model may or may not use correctly
Retrieval changes the model's *inputs*, not its nature. A retrieved document can still be ignored, misread, or contradicted.
2. B - Nothing reliable - it may be reproducing a common pattern that happens to match
A plausible answer is not evidence of a source. Without retrieval and citation you cannot tell recall from coincidence.
3. B - Answering from a defined, changing document set with citations
RAG shines when answers must trace to specific, updatable sources. It is not a calculator and not a permissions system.
4. A - When the knowledge changes frequently and must be cited
Fine-tuning bakes knowledge in and is expensive to update. Retrieval keeps knowledge external, current, and attributable.
5. B - Assuming retrieval quality is fine and blaming the model for bad answers
Most bad RAG answers are retrieval failures. If the right passage never arrived, no prompt can rescue the answer.
Practical Check
Apply this module to your own work: complete the module activity for *Why Normal Chat Is Not the Same as Knowledge Retrieval*, then write one sentence naming what your result shows and one naming what it does not.
Strong Answer Pattern
A strong answer names the task, the evidence used, the check performed, and the remaining limitation. It avoids "proved", "guaranteed", or "always" unless the evidence genuinely supports it.