Final Assessment

AI Classroom Toolkit — Final Assessment

Scenario-based questions drawn from across all modules. Pass mark 80%. Attempt again after focused revision if needed — the toolkit matters more than the first try.

1. An AI tool produces a fluent, confident explanation containing a factual error. What made this possible?

Generative AI predicts likely text and does not verify facts, so confident output can be wrong.The tool was low quality; better tools do not make factual errors.The teacher's prompt was too short, which forces errors.Errors only occur when the topic is not in the syllabus.

2. Which prompt respects the privacy red lines?

"Draft an encouraging remark for a middle-school student whose consistency is improving.""Write a remark for Aarav Sharma of 7B who scored 43%.""Suggest strategies for the girl in my class with dyslexia, Meera.""Summarise this counselling note about a student's family for the principal."

3. In the SATHI framework, the H slot exists to:

Set the human boundaries: what must not go in or come out, and what the teacher will verify.Specify how many questions the output should contain.Tell the AI which language to answer in.Make the prompt polite so the output improves.

4. A teacher wants a lesson plan the class can actually run. Which detail set matters most in the prompt?

Outcome wording, prior knowledge, period length and available materials.The topic name and a request to make it excellent.The name of the textbook publisher only.A long list of adjectives describing good teaching.

5. Why must a teacher solve every item of an AI-generated test before using its answer key?

AI keys can be confidently wrong, and a wrong key fails students unfairly and silently.It is only necessary for essay-type questions.Solving items makes the test longer and harder.The key is always correct; checking is for formatting.

6. A 'support version' of a worksheet comes back with the analysis question removed and a babyish tone. The correct response is:

Iterate: keep the scaffold, restore the analysis question, and demand the same respectful tone as the standard version.Accept it — easier versions should remove hard thinking.Give this version only to the weakest students, labelled accordingly.Abandon differentiation as too risky with AI.

7. Many students consistently answer 1/2 + 1/3 = 2/5. What is this, and what does it need?

A misconception — a consistent wrong model needing confrontation with a counterexample and targeted rebuilding.Carelessness — more practice of the same kind will fix it.A sign these students cannot learn fractions.An AI hallucination in the marking.

8. For an inquiry project, where should AI assistance stop?

After the stimulus (scenario, data, open questions) — before students investigate and conclude.After it has drafted model conclusions for students to check.Nowhere — AI should support every stage equally.Before the project — AI should not be used for inquiry at all.

9. Which administrative task is RED — never for an AI tool?

Summarising a counselling note about a student's family situation.Drafting the sports-day notice for parents.Creating a staff-meeting agenda from five topics.Formatting a classroom supplies inventory.

10. A Hindi translation of a worksheet reads correctly but far more formally than the simple English original. This is:

Register inflation — a translation failure that silently raises the reading level; specify everyday register and re-check.Correct behaviour — Hindi should always be more formal.Acceptable as long as the facts are preserved.Proof that bilingual materials should not be generated.

11. An AI-generated passage cites 'a 2019 NCERT study'. Before using the passage, the teacher must:

Look up the cited study independently — invented citations look perfectly formatted.Ask the same AI whether the study is real.Trust it — NCERT citations are always real.Remove all citations so none can be wrong.

12. The capstone toolkit requires a signed declaration. What does it certify?

No identifiable student data was used in building or testing the toolkit, and none will be entered in its use.The teacher wrote every prompt without any AI assistance.The toolkit is approved by the education board.The AI outputs are guaranteed error-free.