Creating Learning Resources with Generative AI

A ten-step creation process

1. Define the learning objective, audience and intended learner action.
2. Identify trusted source material and non-negotiable content.
3. Decide what role AI will play: brainstorm, transform, generate variants, illustrate, check or format.
4. Write a prompt that includes context, constraints, evidence sources, examples and output criteria.
5. Generate a limited first draft rather than a large quantity of unreviewed material.
6. Verify every factual claim, quotation, reference, calculation and representation.
7. Adapt the draft for pedagogy, learner readiness, language, culture and local context.
8. Check accessibility, bias, privacy, copyright and permissions.
9. Pilot the resource with a colleague or small learner group.
10. Document the version, human edits, evidence basis and intended conditions of use.

Appropriate uses

• Generating alternative examples at different levels of complexity.
• Rewriting teacher-authored text for clarity while retaining the original meaning.
• Creating question variants for human review.
• Producing a first storyboard or visual brief.
• Suggesting misconceptions, discussion prompts or counterexamples to investigate.
• Converting an existing resource into a checklist, summary or differentiated format.

High-risk uses

• Generating factual content without source verification.
• Uploading personal, sensitive, confidential or protected information.
• Using generated references without checking that the sources exist and support the claim.
• Using generated images or text without considering copyright, attribution and cultural representation.
• Allowing AI to make consequential judgements about learners.
• Replacing professional judgement with a generic generated lesson sequence.

Verification protocol

Accuracy — can every claim be confirmed through a trusted source?

Alignment — does the resource serve the stated learning objective and curriculum context?

Pedagogy — what will learners think, decide, practise, explain or produce?

Bias and representation — whose perspectives are present, absent or stereotyped?

Accessibility — can learners perceive, navigate and understand the resource through appropriate alternatives?

Privacy — has any sensitive or identifiable information been disclosed?

Copyright — are source materials, images, text and generated derivatives used appropriately?

Transparency — should users be told that AI contributed to the resource?

Evidence and policy basis

UNESCO's guidance advocates a human-centred approach to generative AI in education and highlights the need for regulation, human capacity, data privacy, ethical use and critical validation. This guide must be checked against current MOE policies, approved tools and security requirements before public release.