Module 10 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. The first design decision for a knowledge assistant should be…
- A. Which vector database
- B. Which questions it must answer, for whom, and what a wrong answer would cost
- C. Which embedding model
- D. The UI framework
2. Where should a human review step sit in a high-consequence workflow?
- A. Nowhere
- B. Between the drafted answer and any action or external communication
- C. Before retrieval
- D. After the action
3. A stop condition for a knowledge assistant is…
- A. A rate limit
- B. A defined situation where it must not answer and must escalate
- C. A timeout
- D. A cost cap
4. Monitoring after launch should track…
- A. Only uptime
- B. Retrieval quality, refusal rate, citation accuracy, and user-reported errors
- C. Only token spend
- D. Only latency
5. Users should be told…
- A. Nothing about how it works
- B. That answers come from a defined document set, may be incomplete, and how to verify or escalate
- C. Only the model name
- D. That it is always accurate
---
Answer Key and Explanations
Check these only after attempting every question.
1. B - Which questions it must answer, for whom, and what a wrong answer would cost
Scope and consequence determine every downstream choice, including how much verification is warranted.
2. B - Between the drafted answer and any action or external communication
Review must occur while it can still prevent the consequence, not after the message has been sent.
3. B - A defined situation where it must not answer and must escalate
Naming in advance where the system must stop is what turns a demo into a deployable workflow.
4. B - Retrieval quality, refusal rate, citation accuracy, and user-reported errors
Corpora and questions drift; behavioural metrics catch degradation that infrastructure metrics miss.
5. B - That answers come from a defined document set, may be incomplete, and how to verify or escalate
Calibrated trust is the goal. Users who know the boundaries use the system more safely and more effectively.
Practical Check
Apply this module to your own work: complete the module activity for *Designing a Knowledge Assistant Workflow*, 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.