A completed project product
AI-application design — completed proposal: Practice Check. Input: a learner’s answer to one teacher-approved practice question, with no name or private record. Output: a suggested hint and a reason to check, not a final ability label. Human role: the teacher approves the question and reviews doubtful feedback; the learner can report an error. Privacy rule: do not store unnecessary personal information. Failure case: if an answer is missing because of a connection problem, ask the learner to retry rather than infer low ability. Final product: an input–decision–output diagram, a worked failure example and a short presentation. No working software or claimed test result is implied.
Completed paper workflow: teacher-approved question → anonymous learner attempt → check whether an answer was actually saved → if missing, request a retry; if present, offer a tentative hint → learner explanation and teacher review. Worked failure example: the learner selects “on” for “Success depends ___ regular use”, but the connection fails before saving. The saved blank is missing evidence, not an incorrect preposition. The output should request another submission, not assign low ability. This is a worked design example, not an observation from a deployed application.