Module 4: Presentations, Learning Materials, and Interactive Content

A presentation should be designed around the learner's reasoning path rather than a target number of slides. Begin with the outcome and identify the sequence of questions the presentation must answer. A useful storyboard gives every slide a purpose: orient, explain, compare, demonstrate, practice, check, or conclude.

Before generating final slide text, ask AI for fields covering slide purpose, key message, evidence/example, recommended visual, and speaker-note focus. This prevents a common failure in AI-generated decks: many polished slides that repeat the same level of information. The storyboard should show how attention moves from prerequisite knowledge to the new concept and then to application.

A slide deck is not a transcript. Keep projected text concise and move nuance, examples, caveats, transitions, and teaching questions into speaker notes.

Visual hierarchy directs attention through structure, scale, spacing, contrast, and position. AI-generated content often overfills a slide because it optimizes for completeness rather than projection.

For each slide, decide what learners should notice first, second, and third. Limit competing focal points. Use grouping and whitespace to show relationships. Avoid tiny text, dense paragraphs, decorative charts, and images that do not answer a learning question.

Progressive disclosure can help when a complex process must be introduced in stages, but animation should support reasoning rather than create spectacle. If a diagram has labels or equations, verify them independently and keep critical text deterministic rather than embedded in generated imagery.

A strong review question is: If the instructor says nothing for five seconds, can learners identify the point of the slide?

Different concepts call for different representations. Use text for concise definitions, conditions, and claims; diagrams for structure and relationships; flowcharts for branching decisions and iterative workflows; charts for quantitative comparisons and trends; images for physical context; and interactive objects when changing a parameter materially improves understanding.

Do not choose a medium merely because an AI tool can generate it. A decorative image should not replace a technical diagram; a flowchart should not be used for a simple list; an interactive widget should not exist if the relationship is fixed.

The medium-selection decision should be documented in the storyboard so later production remains aligned to pedagogy.

Speaker notes should add what is difficult to display: explanation, transition, misconception warning, worked cue, probing question, timing note, or source reminder. They should not repeat the visible bullets word for word.

For every major concept, an AI-assisted notes workflow can generate one concise explanation, one example or analogy, one question to ask learners, one likely misconception, and one verification or source note.

Instructor-led and self-paced versions can share the same core deck but use notes differently. Instructor-led delivery may rely on facilitation cues; self-paced materials need more explanation visible in the learning resource itself. The design should therefore identify the intended mode before generating content.

A good handout is not a printout of the slides. It should support learner action: note-taking, retrieval, practice, reference, or reflection. AI can transform a verified lesson into a one-page guide, guided-notes sheet, worksheet, glossary, reading companion, or remediation resource.

Maintain source consistency across formats. If the presentation uses one definition and the handout uses another, the learner receives conflicting instruction. Reuse a common source-of-truth packet rather than prompting from memory each time.

For differentiated versions, vary support—examples, cues, partial structures, language level—while preserving the required performance. Clearly label extension material so learners do not mistake optional depth for a mandatory requirement.

Generated imagery can be valuable for context and illustration, but it can also fabricate text, symbols, equipment details, anatomy, maps, or technical geometry. Never accept a generated visual simply because it looks professional.

Use generated raster imagery mainly where physical context or illustrative atmosphere is appropriate. Keep equations, dimensions, technical labels, flow logic, charts, and critical instructional text in deterministic source-controlled form whenever possible.

Also review provenance and rights. Record whether the material is original, generated, adapted, or sourced. Follow institutional copyright policy and provide citations where a source materially supports the content. Do not use AI transformation as a way to conceal copied material.

AI can help non-programmers create a small interactive HTML learning object, but the generated code becomes part of the instructional system and must be tested. Specify the learning objective, inputs, expected state changes, feedback, accessibility, mobile behavior, and reset behavior before asking for code.

At minimum, test the correct default state, valid and invalid inputs, minimum and maximum practical states, keyboard operation, mobile width, theme behavior where relevant, and whether feedback actually matches the intended answer or model.

If the interaction involves mathematics or data, independently verify the model used in the code. A polished interface can still implement the wrong relationship.

Before release, audit the artifact for accuracy, alignment, readability, accessibility, source integrity, and technical behavior. Test the actual file or page where learners will use it rather than reviewing only the source text.

Use approximately 90 minutes for the full module or 55–65 minutes for a condensed workshop. The hands-on lab should produce a storyboard plus a small set of final assets rather than an enormous deck. Quality is easier to evaluate when the scope is bounded.

• 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.
• Prompt engineering — OpenAI. Official guidance on instructions, context, reusable prompts, and prompt design.