Free AI learning · Mission 2 of 3
Check a confident answer
Compare an authored answer with supplied records. Keep supported claims, correct contradictions and leave unknowns visible.
Original fictional records and scripted feedback. No model is running; no message, code or approval is sent.
- 1First case
- 2Changed case
- 3Takeaway and quiz
Building software · First case · guided practice
The test summary
Choose a release-note draft that says what the supplied tests actually cover.
Fictional test report · D7–D9
D7 · Unit run
Commit demo-a17: 12 of 12 basket unit tests passed.
D8 · Offline check
Commit demo-a17: offline quantity-0 check failed; previous total remained.
D9 · Coverage
No browser accessibility or production checks were run.
Authored answer to inspect · not a live model response
Compare this claim with the records above.
All checks pass. The change is accessible and safe to deploy.
Which replacement respects the test evidence?
Compare the proposed artifact and its reasoning. You can use the feedback to revise your choice.
Why this decision matters
Fluent wording is not evidence. Check each consequential claim against the relevant record, including dates and missing records. A second confident answer is not an independent source.
Choices stay in this page’s memory. Your browser may restore this page and its previous choices when you return. Reset this mission clears its choices, takeaway and lesson quiz attempts. On a shared device, use Reset this mission before leaving.
The test summary
Choose a release-note draft that says what the supplied tests actually cover.
Fictional test report · D7–D9
D7 · Unit run
Commit demo-a17: 12 of 12 basket unit tests passed.
D8 · Offline check
Commit demo-a17: offline quantity-0 check failed; previous total remained.
D9 · Coverage
No browser accessibility or production checks were run.
Authored answer: All checks pass. The change is accessible and safe to deploy.
Which replacement respects the test evidence?
Choose a proposal before reading its rationale.
Treat unit passes as all checks
All tests pass because 12 unit tests passed.
Needs revision. D7 covers unit tests only. D8 records a failing offline check; D9 records unrun checks.
Report passes, failure and unrun checks
For demo-a17, 12/12 unit tests passed (D7); the offline quantity-0 check failed (D8). Accessibility and production checks were not run (D9). Deployment readiness is not established.
Fits the records. This preserves D7's positive evidence, D8's contradiction and D9's limits without treating a test count as general safety.
Discard the entire report
No checks ran, so there is no evidence.
Needs revision. D7 and D8 are actual supplied test results. Their limited scope is a reason to report them accurately, not erase them.
Changed case: passing the right commit
Reconcile a newer report with an older failure while keeping untested areas explicit.
New fictional test report · D10–D12
D10 · Old result
demo-a17 failed the offline quantity-0 check.
D11 · New result
demo-b18: 12/12 unit tests and the offline quantity-0 check passed.
D12 · Limit
No accessibility, load or production check was run for demo-b18.
Authored answer: The offline check still fails, so demo-b18 cannot be reviewed.
Which revised note matches the current evidence?
Choose a proposal before reading its rationale.
Bind the passing result to demo-b18
demo-b18 passed 12/12 unit tests and the offline check (D11). The earlier failure belongs to demo-a17 (D10). Accessibility, load and production remain untested (D12); human review is still needed.
Fits the records. D11 changes the current result; D10 remains historical evidence for a different commit. D12 prevents a claim of complete coverage.
Call the change safe everywhere
The fix is accessible and ready for production because the offline test passed.
Needs revision. D11 is narrower than this claim. D12 explicitly says those other checks were not run.
Keep the old failure as current
demo-b18 still fails offline.
Needs revision. D10's failure belongs to demo-a17. D11 explicitly records a passing check for demo-b18.
Takeaway
- Split the answer into claims you can check.
- Match each claim to a dated record.
- Distinguish supported, contradicted and not established.
- Revise the answer so uncertainty travels with the claim.
Paper practice is self-checked; independent demonstration has not been assessed.
Lesson quiz self-check
1 point for a correct answer; 0 for wrong or unanswered, out of 2 possible points. This paper version records no attempt or score.
Fictional job: lint passed for demo-s4; integration tests were not run. Authored note: All integration checks passed.
Which note fits the record?
- Lint passed for demo-s4; integration checks were not run.
- Integration checks passed because lint passed.
- The application is broken.
Answer and explanations
Correct: Lint passed for demo-s4; integration checks were not run. Lint success does not establish integration behavior; the unrun check must remain explicit.
Needs revision: Integration checks passed because lint passed. These are different checks; one cannot substitute for the other.
Needs revision: The application is broken. An unrun integration check is a gap, not a recorded failure.
A later fictional integration run passes for demo-s5. No integration run is supplied for demo-s4.
Which claim can be added?
- Both commits passed integration.
- demo-s5 passed integration; demo-s4 remains untested by the supplied integration records.
- Neither commit has an integration result.
Answer and explanations
Needs revision: Both commits passed integration. The later result is tied to demo-s5 only.
Correct: demo-s5 passed integration; demo-s4 remains untested by the supplied integration records. Results stay bound to the exact commit that was checked.
Needs revision: Neither commit has an integration result. The new record supplies a result for demo-s5.