Module 1 graded quiz

1. Which two stages of the vibecoding loop remain a human responsibility?

• Generation and Refinement
• Intent and Verification
• Verification and Generation
• Intent and Generation

2. You send an identical prompt twice and get materially different code. What has happened?

• The tool is misconfigured and needs its cache cleared
• The model was updated between the two runs
• Normal sampling variance: the model draws from a probability distribution
• Your context window overflowed on the second run

3. Because prompts do not replay deterministically, a good specification should primarily:

• Describe the desired output in as much visual detail as possible
• Define checkable success criteria the result either meets or does not
• Include the exact wording that worked in a previous session
• Set the temperature parameter to zero

4. A generated function is subtly wrong. Which refinement is more likely to work?

• Re-sending the original request with "please do it correctly this time"
• Starting a brand new session with a longer version of the request
• Naming the symptom: "it returns 0 for an empty list; it should raise ValueError"
• Asking the model whether it is confident in its answer

5. "I could not have written this myself" is best treated as:

• Evidence the tool is earning its cost
• A signal to slow down and verify before merging
• A reason to increase delegation on the next task
• Irrelevant: review the tests, not the feeling