Designing repeatable group AI assessments
How saved questionnaire rounds, private feedback and aggregate charts help a mixed-experience group reflect on AI practice and revisit what changed.
The problem
A course host needs to understand a mixed-experience group and revisit progress later. Repeating a one-off technical quiz would re-ask stable background questions and lose the earlier answers.
The approach
Each session saves an immutable question snapshot and completed answers. Follow-up rounds keep initial programming background while asking current practice and knowledge again.
Private feedback belongs to the participant. Public breakdowns require at least three responses; signed-in hosts can preview with two. Comparisons between matched rounds retain their own minimums.
Course introduction, day recap and course recap use ungraded questions. Hosts can reveal word clouds, selection charts and written responses in a presentation, and reopen saved sessions.
What the implementation enables
The host can distinguish initial background from current practice, revisit earlier rounds and discuss aggregate changes. Recaps capture takeaways, applications and project direction without a grade.
How it is checked
Repository and API tests cover immutable saves, retry behavior, question snapshots and response thresholds. Component tests cover answering, saved history and the presentation controls.
Limits and tradeoffs
This is a learning diagnostic, not a certification. Matching uses an anonymous browser identity; a new device or cleared storage can lose the match. Free-text answers should omit identifying details.