Using Generative AI for Curriculum Design
This guide proposes ways to use generative AI to organise supplied requirements, compare alternatives, and draft planning artefacts. Research on AI-assisted inquiry-based lesson planning provides a related example at session level, rather than validation of a whole-curriculum workflow.1 Curriculum quality still depends on constructive alignment, disciplinary judgment, institutional requirements, and knowledge of the students and program.
Establish the Purpose of the Course
Before using AI, clarify the course purpose and its place in the wider program. Constructive alignment connects intended outcomes with teaching activities and assessment evidence.2
Resolve these questions with the course team before prompting:
- What do my students need to learn?
- How does this course connect to larger program goals, learning outcomes, or standards?
- What outcomes, knowledge, or skills will have the greatest impact on student growth?
Provide approved requirements and context, then treat any generated structure as a draft to test against them.
Example prompt:
Using the supplied program outcomes and constraints, propose two possible structures for an advanced biology course focused on critical analysis of research papers. Map each unit and assessment to an outcome, flag missing information, and do not invent institutional requirements.
Begin with Program Requirements and Learning Outcomes
Effective curriculum design begins with the approved program outcomes, qualification framework, professional or accreditation requirements, and local policies. AI can help organise supplied requirements into candidate course outcomes, but it should not invent or interpret authoritative requirements on its own.
Review proposed outcomes for clarity, level, assessability, and coverage. Bloom’s Taxonomy can help describe cognitive demand, but higher levels are not automatically more appropriate; the level must fit the program and learner stage.
Example prompt:
Using the supplied program outcomes, review these course outcomes for clarity, assessability, level, and coverage. Propose revisions without introducing requirements absent from the source. Return an outcome-to-program mapping and flag gaps.
Provide Course Context
Provide the course title, credits or expected workload, learner level, duration, prerequisites, approved topics, assessment constraints, available resources, and accessibility requirements. Ask for an outcome-to-unit map before a detailed weekly schedule so gaps and unnecessary topics remain visible.
Plan progression, workload, and evidence
Map outcomes across units so that prerequisites come before dependent tasks, important capabilities receive repeated practice, and assessments have adequate preparation. Ask AI to expose gaps and duplication rather than simply allocate a topic to every week. Check contact hours, independent study, assessment load, and available staff time against the approved course workload.
Using the supplied course context and outcomes, return a unit-level map with:
- prerequisites and the outcome practised;
- the main student activity and observable evidence;
- the assessment or formative check;
- estimated contact and independent-study time; and
- gaps, repeated content, or workload peaks for the course team to resolve.
Do not change approved outcomes or invent accreditation requirements. Label
workload estimates for human review.Once the map is approved, use Planning lessons for session design, Creating course materials for production, and Creating exam questions for items.
Enhance Interactive Learning Experiences
Ask AI for candidate discussions, cases, simulations, collaborative problems, or projects. Request student actions and observable evidence, not merely “engaging activities”. Reject novelty that does not serve an outcome, cannot be completed with available resources, or creates avoidable participation barriers.
Example prompt:
Using the supplied outcome and policy sources, propose a 30-minute role-play in which students compare competing environmental-policy perspectives. State what students do, what evidence of learning is observable, what preparation is required, and an accessible non-role-play alternative.
Worked alignment example
Suppose an outcome says: “Evaluate competing environmental-policy proposals using economic, distributional, and implementation evidence.” Instead of asking for generic engaging activities, supply the outcome and relevant course sources:
Propose three activities for this outcome: one preparation task, one 30-minute
class activity, and one formative check.
For each activity, state:
- what students do;
- which part of the outcome they practise;
- what evidence of learning the educator can observe;
- required prior knowledge and materials;
- accessibility or participation barriers; and
- one feasible alternative.
Use only the supplied course sources for factual content. Do not require an AI
tool unless AI literacy is part of the outcome.A suitable result might pair a short evidence-comparison task with a stakeholder negotiation and an individual written justification. The educator should reject ideas that are merely entertaining, cannot be assessed, duplicate another activity, or cannot be completed with the available time and access.
Before finalising the curriculum, build a matrix linking every course outcome to learning activities and assessment evidence. Gaps in either column indicate that the curriculum is not yet constructively aligned.
Use AI output to expose choices, not to approve the curriculum. The responsible course team must verify requirements, workload, progression, assessment, accessibility, and coherence and remain able to justify the final design.
Map progression across a programme
A course-level map can miss a programme-wide gap: students may repeatedly use AI-supported explanations without ever demonstrating the relevant capability independently. Map where the capability and responsible tool use are introduced, practised, and assessed. The following is an illustrative three-stage sequence, not an accreditation requirement or a validated curriculum.
| Stage | Independent capability | AI-related learning | Evidence and handover |
|---|---|---|---|
| Foundation | Interpret a claim and a simple dataset | Recognise unsupported output and check a source | A fresh claim check and denominator calculation; identify prerequisites for the next course |
| Intermediate | Compare methods and justify an analysis | Critique an AI proposal and document a consequential correction | A reproducible analysis, justified revision, and independent new-case check |
| Advanced | Defend a professional decision under uncertainty | Select, test, or reject an AI workflow | A project decision record, limitations, and an accessible structured defence |
Use data analysis, entrepreneurial learning, and the worked assessment designs as candidate activities. Adapt the progression to the actual discipline; not every programme needs the same technical tasks or required AI use.
For each programme outcome, identify the course responsible for introduction, later practice, and sufficient assessment evidence. Confirm that a prerequisite is taught before another course relies on it. Check cumulative student workload, staff marking capacity, and whether several courses require incompatible tools or duplicate disclosures. Reuse SAGE’s competency pathway as a planning reference rather than adding a separate AI exercise to every unit.
The programme team should review gaps and redundant tasks together, using proportionate, authorised assessment evidence. Do not create a central archive of student chats to track progression. Retain independent checks for capabilities graduates must demonstrate without AI, with the usual approved accommodations. Record who owns the map and revisit it when outcomes, courses, or tools change.
Further Reading
- Ayres, A. T. (2025, February 25). A curriculum supervisor’s guide to AI-assisted lesson planning. Edutopia; George Lucas Educational Foundation. https://www.edutopia.org/article/ai-generated-lesson-plans
References & Footnotes
Footnotes
-
Moundridou, M., Matzakos, N., & Doukakis, S. (2024). Generative AI tools as educators’ assistants: Designing and implementing inquiry-based lesson plans. Computers and Education: Artificial Intelligence, 7, 100277. https://doi.org/10.1016/j.caeai.2024.100277 ↩
-
Biggs, J. (1996). Enhancing teaching through constructive alignment. Higher Education, 32(3), 347–364. https://doi.org/10.1007/BF00138871 ↩