Final Assessment

AI Leadership for School Coordinators and Institution Heads — Final Assessment

Forty questions across the orientation and ten modules — including scenario, policy-interpretation, risk-classification and vendor-comparison items. Pass mark 80%. Every question explains why answers are correct or incorrect. Attempt again after focused remediation if needed.

1. What must an institutional AI leadership programme enable, beyond tool use?

Governing AI adoption at institutional level.Only using AI tools personally.Coding AI systems.Marketing the school.

2. Before adopting AI, what should a leader assess first?

Multi-domain institutional readiness.The cheapest available tool.Competitor schools' choices.Nothing — just start.

3. In a RACI, how many people are Accountable for a decision?

Exactly one.As many as possible.Zero.The vendor.

4. Who remains accountable for an AI outcome when the tool is a third-party product?

The institution.The vendor alone.No one.The students.

5. What is the real goal of teacher AI capacity building?

Professional judgement across tools.Mastery of one product.Fastest adoption.Memorising prompts.

6. Why is mandating unprepared AI adoption a governance risk?

It raises data and safety incident risk.It has no downside.It speeds safe adoption.It only affects morale.

7. How should responsible AI principles be applied?

As decision criteria for each use.As decorative slogans.Only after adoption.Never in practice.

8. What does proportionality require before an intrusive AI tool is adopted?

Evidenced benefit justifying the intrusiveness.Only vendor popularity.A low price.Nothing.

9. Which AI use case belongs in the prohibited tier for schools?

Emotion recognition and behaviour prediction of students.Drafting a lesson plan.Translating a notice.Reviewed worksheet generation.

10. What is the clearest red line in a school acceptable-use policy?

No identifiable student data in public AI tools.Banning all AI use.Requiring AI for every task.Nothing needs prohibiting.

11. Why must prompt data be treated as potentially sensitive?

It may contain sensitive data and be stored/trained on.Prompts never leave the tool.Prompts are not data.Only files are data.

12. What does data minimisation require?

Only what the purpose genuinely needs.As much as possible.Data for unknown future uses.Nothing at all, ever.

13. How should a course treat legal claims about children's data?

Separate legal from policy; recommend legal review.Present it as binding advice.Claim certified compliance.Ignore the law.

14. Why is shadow AI a governance concern?

It bypasses review and creates untracked risk.It is always safe.It is official policy.It has no privacy impact.

15. What is the correct first action when student data is exposed to a public AI tool?

Contain and preserve evidence.Delete records quietly.Do nothing.Blame the vendor first.

16. Why are low-bandwidth and shared-device support requirements, not extras?

Otherwise adoption becomes an equity problem.Bandwidth never matters.Only staff devices matter.They are optional nice-to-haves.

17. Which vendor criteria most protect student data?

Data ownership, no-training and exit provisions.The number of features.The demo quality.The brand name.

18. Why can a 'free' public AI tool cost more overall than a paid platform?

It hides support, risk and migration costs.Free is always cheapest overall.It has no hidden costs.Cost never depends on support.

19. What should drive an AI vendor decision?

A weighted, evidence-based scorecard.The flashiest demo.Marketing claims.The first vendor to call.

20. What is the constant that must accompany AI use in teaching?

Teacher review before output reaches students.No review is needed.Vendor sign-off.A spelling check only.

21. Can an AI-detection score be the sole basis for a misconduct finding?

No — never sole evidence.Yes, it is definitive.Yes, if the score is high.Yes, for take-home work.

22. What is the most durable response to AI in assessment?

Authentic assessment redesign.More detection software.Banning technology.Invasive monitoring.

23. What distinguishes proportionate monitoring from surveillance?

Purpose-limited, aggregated and transparent.Secretly tracking every individual.Collecting everything.Profiling teachers.

24. Why is rising AI usage an insufficient measure of value?

Usage is activity, not improved outcomes.Usage always proves value.Outcomes cannot be measured.Only usage matters.

25. What is the correct focus of effective incident management?

Root-cause fixes to the process.Blaming an individual.Hiding the incident.Ignoring minor incidents.

26. Why state where AI is NOT used when communicating to parents?

It reassures humans decide about children.It is unnecessary.It hides AI use.It confuses parents.

27. Why must an AI adoption include a non-AI alternative?

So students without access are not excluded.Alternatives are a waste.Everyone has equal access.To slow adoption.

28. Why consult students and parents before adoption?

It builds partnership and trust.It slows everything needlessly.Opinions do not matter.Only the principal decides.

29. Where should an institution start prioritising AI use cases?

Low-risk, high-impact quick wins.The riskiest use case.Everything simultaneously.Whatever is cheapest.

30. Why must a ninety-day pilot include a stop criterion?

To stop weak or risky pilots by design.Stop criteria are unnecessary.Pilots never stop.To please vendors.

31. What is the purpose of a decision gate between roadmap phases?

To advance only when a phase's criteria are met.To scale automatically.To skip evaluation.To hide budgets.

32. How should a capstone answer a board's 'why fund uncertain AI?' question?

With bounded pilots, controls and evidence.With promises of transformation.By asking for an open budget.By avoiding the question.

33. What must the capstone plan acknowledge about its content?

No personal student data; not legal advice.That it is government-approved.That it guarantees outcomes.Nothing.

34. Should this course or its certificate claim government or board accreditation?

No — it is a learning credential only.Yes, always.Yes, government-recognised.Yes, a statutory licence.

35. What must every module require of the learner at least once?

A realistic leadership decision.Only reading and clicking next.A payment.Nothing active.

36. How should factual, legal or regulatory statements be handled?

Sourced, dated and never fabricated.Invented if convenient.Presented as certain law everywhere.Without any source.

37. Should the embedded AI assistant answer graded assessment questions?

No — it must not answer graded questions.Yes, always.Yes, for hard questions.Yes, if asked.

38. What is the passing score for the final knowledge assessment?

80%.50%.100%.No pass mark.

39. What is the recommended capstone passing score?

70% on the 100-point rubric.Any score.100% only.It is not assessed.

40. What is the practical output a learner produces by completing the programme?

A complete institutional AI leadership portfolio.A single certificate and nothing else.A coding project.A marketing brochure.