Module 06 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. 'Grounded' means each material claim…
- A. Sounds confident
- B. Is supported by a retrieved passage the reader can verify
- C. Has a citation attached
- D. Came from the model
2. Instructing the chain to answer only from context is…
- A. A hard guarantee
- B. A useful bias that reduces but does not eliminate unsupported additions
- C. Unnecessary
- D. Harmful
3. When retrieved context does not contain the answer, the chain should…
- A. Answer from general knowledge
- B. State that the sources do not cover it
- C. Return the closest chunk
- D. Retry silently
4. Citations should reference…
- A. The whole document
- B. The specific passage used, with a path back to the source
- C. The vector id
- D. The model version
5. Stuffing more retrieved documents into the prompt to improve quality often…
- A. Always works
- B. Backfires - relevant content gets diluted among distractors
- C. Reduces cost
- D. Improves citations
---
Answer Key and Explanations
Check these only after attempting every question.
1. B - Is supported by a retrieved passage the reader can verify
An attached citation that does not support the claim is worse than none - it manufactures trust.
2. B - A useful bias that reduces but does not eliminate unsupported additions
Prompt instructions shape behaviour; they do not enforce it. Verification stays in the workflow.
3. B - State that the sources do not cover it
Falling back to parametric knowledge in a grounded assistant produces confident, uncitable claims.
4. B - The specific passage used, with a path back to the source
Verification must be cheap or it will not happen.
5. B - Backfires - relevant content gets diluted among distractors
Precision of context matters more than volume; re-ranking beats raw quantity.
Practical Check
Apply this module to your own work: complete the module activity for *Retrieval Chains with Grounded Responses*, 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.