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. Calling a model three times with the same prompt…
- A. returns the same string
- B. returns three different, possibly all correct, strings
- C. raises on the second call
- D. returns cached output
2. Which failure charges you and leaves you unable to tell whether it worked?
- A. 429 rate limit
- B. a timeout
- C. 401 unauthorised
- D. a validation error
3. In a well-structured AI pipeline, the number of non-deterministic steps is…
- A. one per component
- B. one, with validation immediately after
- C. as many as needed
- D. zero
4. A guarantee a script has and an AI app does not is…
- A. it can be version controlled
- B. its behaviour is stable over time
- C. it can be logged
- D. it can be deployed
5. Each line of a must-never list needs…
- A. a priority
- B. an enforcement that is not a prompt instruction
- C. an owner
- D. a severity rating
---
Answer Key and Explanations
Check these only after attempting every question.
1. B - returns three different, possibly all correct, strings
Design tests, caches and comparisons around variation rather than trying to suppress it.
2. B - a timeout
The call may have succeeded upstream. Anything with a side effect behind it needs an idempotency key.
3. B - one, with validation immediately after
Past the validation you have a known shape and known values, and everything downstream is testable.
4. B - its behaviour is stable over time
A hosted model can change behaviour with no commit on your side, which is why evals run on a schedule.
5. B - an enforcement that is not a prompt instruction
A prompt is read by a component that has already been handed the thing you want withheld.
Practical Check
Apply this module to your own work: complete the module activity for *What Makes an AI App Different from a Script*, 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.