Skip to lesson content
Free course

AI Foundations / Module 7 / M07-U05 · 20-30 minutes

M07-U05 · 20-30 minutes

Test, Diagnose, and Revise

Work through the explanation, apply it to the example, and complete the quick check before continuing.

Test and revision log showing safe test cases, expected and actual results, defect, likely cause, targeted change, retest result, and a decision to change method or reject the workflow.
Run safe normal, missing-information, and conflicting-information tests where relevant; record one meaningful defect, make one targeted change, and retest before deciding whether the workflow is suitable.
Text description of the visual

The figure shows a capstone test and revision log. The safe test set includes a normal case, a missing-information case, and a conflicting-information case where relevant. The revision log records expected result, actual result, defect found, likely cause, targeted change, and retest result. A defect example area lists unsupported claim, omitted need, wrong tool, unclear handoff, inaccessible format, and missing stop condition. The decision area says that if retests remain unsuitable, learners should change the method or reject the workflow. The final reminder says that the test must expose at least one meaningful defect, and that honest failure is stronger than hidden defects.

Run the workflow on safe material. Include normal, missing-information, and conflicting-information cases where relevant.

Record expected result, actual result, defect, likely cause, targeted change, and retest result. Do not change everything at once.

Your test must expose at least one meaningful defect. Examples include an unsupported claim, omitted requirement, wrong tool, unclear handoff, inaccessible format, excessive permission, or missing stop condition.

If repeated tests remain unsuitable, change the method or reject the workflow. Honest failure and a justified stop are stronger than hidden defects.