Module 3: Instructional Design, Planning, and OBE

AI-assisted planning should begin with the actual requirements that govern the course: curriculum standards, program outcomes, course outcomes, syllabus, institutional calendar, contact hours, assessment policy, learner prerequisites, and available resources. The AI should not invent a course structure when those requirements are absent.

A useful source map distinguishes fixed constraints from design choices. A mandated outcome, contact-hour requirement, or approved grading policy is fixed unless the institution changes it. Examples, sequence, activities, scaffolds, and teaching media are design choices that the educator may adapt.

Before generating a plan, ask the system to list each required input as PROVIDED, MISSING, AMBIGUOUS, or NOT REQUIRED. Missing inputs that would materially change the lesson should block finalization. This simple gate prevents polished but noncompliant plans.

A learning outcome should describe observable learner performance rather than an instructor activity or vague intention. “Understand retaining walls” is difficult to assess; “evaluate retaining-wall stability against sliding and overturning for a stated loading case” identifies a performance and evidence.

Use cognitive taxonomies as planning aids, not as rigid formulas. A course can include remembering, explaining, applying, analyzing, evaluating, and creating, but the verb alone does not determine the real cognitive demand. “Calculate” may be routine or highly analytical depending on the problem.

For each outcome, specify the performance, the conditions, the evidence, and the expected level. AI can help repair vague outcomes, but the educator must confirm that the revised outcome still represents the intended curriculum.

Constructive alignment connects intended outcomes → learning experiences → assessment evidence. The key test is not whether an activity is interesting, but whether it prepares learners to demonstrate the intended performance.

Common alignment defects include an outcome that is never assessed; an activity that consumes time but supports no stated outcome; an exam demanding analysis when instruction only practiced recall; a rubric rewarding presentation polish when the outcome is technical reasoning; or excessive assessment of one outcome while another receives almost no evidence.

Ask AI to produce an alignment matrix and flag orphan activities, unassessed outcomes, unsupported assessment demands, and overloaded outcomes. The educator should then correct the smallest structural problem rather than adding more material indiscriminately.

A practical lesson architecture can include prerequisite retrieval, a concise explanation, a worked model, guided practice, independent application, formative checking, and closure. Not every lesson needs every stage in equal proportion, but the sequence should respond to learner readiness.

Scaffolding temporarily reduces task complexity without lowering the essential outcome. Examples include partially completed diagrams, structured prompts, cue cards, worked examples, checklists, or guided questions. Supports should be withdrawn as learners gain competence.

AI can generate differentiated versions of an activity, but ask it to preserve the same essential performance expectation. A scaffolded path should not quietly become a different outcome. For extension learners, increase complexity, transfer, or independence rather than simply adding more questions.

Differentiation can change representation, examples, pacing, support, grouping, or mode of response while preserving core learning expectations. Avoid using AI to make unsupported inferences about a learner's ability, disability, motivation, or background.

Accessibility should be designed from the beginning. For digital materials, consider readable structure, contrast, keyboard access, alt text, captions/transcripts, plain-language instructions, and alternatives to color-only meaning. When AI produces images or diagrams, critical labels should be deterministic and verified rather than trusted as generated pixels.

Ask the planning workflow to identify barriers created by the activity itself: unnecessary reading load, inaccessible file formats, time assumptions, device requirements, or culturally narrow examples. The goal is not identical experiences but equitable access to the intended learning.

For Philippine higher education, CHED's outcomes-based quality-assurance framework remains an important institutional context. CMO No. 46, s. 2012 established outcomes-based and typology-based QA, while CMO No. 37, s. 2012 addressed OBE systems for engineering programs. Program-specific PSGs may add more detailed outcomes and requirements.

Use institutional definitions and approved thresholds rather than inventing a universal attainment rule. A typical evidence chain is Program Outcomes → Course Outcomes → Lesson Outcomes → Learning Activities → Assessment Evidence → Attainment Analysis → Improvement Action → Follow-Up Evidence.

Continuous Quality Improvement is not the act of writing a narrative after grades are submitted. A defensible CQI record identifies evidence, interprets the issue cautiously, records an implemented action, and examines subsequent evidence. AI can help organize the documentation but should not invent interventions or results that never occurred.

In Philippine basic education, DepEd Order No. 003, s. 2026 provides foundational guidance for AI use in teaching, learning, assessment, professional development, and governance. The order emphasizes safe, ethical, pedagogically sound, and human-centred use.

For both K–12 and higher education, the course should teach a common rule: institutional policy and authoritative curriculum requirements override generic AI suggestions. Educators should supply the applicable school or university rules as context rather than assuming a public AI system knows local implementation details.

When adapting a workflow across sectors, preserve the design logic but change the governing sources, learner level, assessment policy, and data-handling requirements.

Allocate 80–100 minutes for the full module or 55–65 minutes for a condensed workshop. A strong instructor demonstration takes one syllabus outcome and shows the progression from weak outcome → measurable outcome → aligned lesson → assessment evidence → CQI-ready mapping.

The participant activity should use a real course or unit whenever possible. The resulting alignment map will later feed the presentation, assessment, grading, and capstone modules.

• Guidance for generative AI in education and research — UNESCO. Human-centred guidance for generative AI in education and research.

• Web Content Accessibility Guidelines (WCAG) 2.2 — W3C. Accessibility guidance for web content and interactive learning materials.

• Artificial Intelligence Risk Management Framework (AI RMF 1.0) — NIST. Framework for governing, mapping, measuring, and managing AI risk.

• 2012 CHED Memorandum Orders — CMO No. 46 and CMO No. 37 — Commission on Higher Education, Philippines, outcomes-based QA and engineering OBE policy listings.

• DepEd Order No. 003, s. 2026 — Foundational Guidelines on AI in Basic Education — Department of Education, Philippines.