AI Guides: UDL and AI
🤖 AI Guides @ JCU
Core principle: Universal Design for Learning (UDL) means providing multiple ways to access content, engage with material, and demonstrate learning—without lowering standards. AI can help provide these pathways, but must preserve challenge and cognitive demand.
1) Representation checks (multiple ways to access content)
Goal: Remove barriers to perception and comprehension without removing complexity.
❑ Text-to-speech is enabled for all written materials
Use: Screen readers, browser read-aloud, or AI voice tools (e.g., Natural Reader, Immersive Reader).
Preserves challenge: Students still process the same complex content, just through audio.
❑ Provide transcripts and captions for all video/audio
Use: AI transcription tools (Panopto auto-captions—reviewed for accuracy).
Preserves challenge: Same content, accessible to students who have hearing issues and those in noisy environments.
❑ Offer visual alternatives to text-heavy explanations
Use: AI-generated diagrams, concept maps, flowcharts (Miro, Canva AI, ChatGPT with DALL-E for simple visuals). Be mindful of the content you share on these platforms, as the data might not be secure. General concepts can be shared to generate diagrams.
Preserves challenge: Diagrams should illustrate relationships, not simplify concepts. Keep cognitive load appropriate.
❑ Check readability and offer vocabulary support—not substitution
Use: AI to define discipline-specific terms in context (glossary links, hover definitions).
Preserves challenge: Do NOT simplify the text itself. Students must engage with academic language; just provide on-demand definitions.
❑ Provide content in multiple modalities when possible
Use: Same concept explained via text reading, mini-lecture video, and worked example.
Preserves challenge: All versions maintain the same depth; students choose their entry point.
2) Engagement checks (multiple ways to sustain motivation and effort)
Goal: Increase relevance, choice, and self-regulation without reducing rigour.
❑ Offer choice in assessment format (where appropriate)
Example: Demonstrate understanding via written report, video presentation, or annotated portfolio—all assessed to the same rubric.
Preserves challenge: Criteria remain consistent; format flexibility reduces anxiety and increases engagement.
❑ Use AI to personalise examples or context
Use: Ask AI to generate discipline-relevant scenarios for different student interests (e.g., medical, business, engineering contexts for the same statistics problem).
Preserves challenge: The underlying problem complexity stays the same; context shifts to increase relevance.
❑ Provide self-paced checkpoints with AI feedback
Use: Low-stakes quizzes or practice problems with AI-generated explanations for wrong answers.
Preserves challenge: Feedback guides reflection, doesn't give answers. Students must still do the cognitive work.
❑ Build in metacognitive prompts
Use: AI chatbot that asks "What strategy did you try?" or "Why do you think that didn't work?" rather than just giving hints.
Preserves challenge: Prompts scaffold thinking process, not content mastery.
❑ Minimise anxiety triggers without lowering expectations
Example: Allow students to submit draft outlines for AI-generated feedback before final submission; grade only the final.
Preserves challenge: Standards for final work remain high; intermediate support reduces panic and improves learning.
3) Action & expression checks (multiple ways to demonstrate learning)
Goal: Let students show what they know through varied means without compromising assessment validity.
❑ Allow assistive writing tools for students with documented needs
Use: Speech-to-text (Google Voice Typing), grammar checkers (Grammarly) for students with dyslexia or motor impairments.
Preserves challenge: Tools assist with mechanics (spelling, typing), not with generating ideas or arguments. Assess content, not transcription.
❑ Permit AI for scaffolding structure, not substituting thought
Example: Students may use AI to generate an essay outline or report template—but must fill it with their own analysis and evidence.
Preserves challenge: Structural support reduces executive function load; intellectual work remains with the student.
❑ Offer alternative output formats for demonstrating mastery
Example: Video explanation instead of a written essay for students with writing disabilities (same rubric: clarity, evidence, reasoning).
Preserves challenge: Mode changes, not the cognitive demand. All formats are assessed against the same learning outcomes.
❑ Use AI for translation support, not content generation
Use: Students whose first language is not English may use AI translation tools to check phrasing—but ideas and arguments must be their own.
Preserves challenge: Language barrier reduced; conceptual rigour maintained. Pair with a viva or oral check if needed.
❑ Provide accessible submission formats
Use: Accept Word, PDF, video, audio, or other formats as appropriate. Ensure your LMS and feedback tools are screen-reader compatible.
Preserves challenge: Submission method doesn't affect content quality or assessment criteria.
4) Over-scaffolding warning signs (and how to fix them)
Warning: AI generates answers instead of guiding the process
Fix: Reframe AI prompts to ask questions, not provide solutions. Example: "What factors should I consider?" not "What is the answer?"
Warning: Simplifying content so much that learning outcomes aren't met
Fix: Keep original texts; add supports like glossaries, guided questions, or pre-reading organisers. Don't substitute easier texts.
Warning: Students skip struggling and jump straight to AI hints
Fix: Require students to attempt the problem first, submit their attempt, then access AI feedback. Build in productive struggle time.
Warning: AI does all organisational or planning work
Fix: Teach planning explicitly; use AI as a check after the student creates an initial plan, not as the plan generator.
Warning: Providing so many supports that the assessment is no longer valid
Fix: Ask: "Does this support help access the task, or does it complete the task?" If the latter, scale back.
5) Quick accessibility audit (5-minute check for any course material)
❑ Can a screen reader navigate this document? (Headings, alt text, logical reading order)
❑ Is there sufficient colour contrast? (Minimum 4.5:1 for text; use WebAIM contrast checker)
❑ Do videos have captions? (Auto-generated is OK if reviewed for accuracy)
❑ Are images, charts, and graphs described with alt text or longer descriptions?
❑ Can students access this on mobile devices? (Check font size, navigation, video players)
❑ Is essential content available in more than one format? (Text, audio, visual)
❑ Are instructions clear and explicit? (Avoid "see above" or "click here"—use descriptive links)
7) Inclusive AI policies (for your course outline)
Consider including language like:
"Assistive technologies and AI tools that support access (such as text-to-speech, grammar checkers, translation aids, or organisational templates) are welcome. These tools should help you engage with the material and demonstrate your learning, not replace your thinking or complete the task for you."
"If you are using accessibility tools as part of a documented support plan, please connect with AccessAbility services or the learning centre to ensure we have appropriate accommodations in place. You do not need to disclose your specific needs to me, but I am happy to discuss how assessment formats can be adjusted."
"If you are unsure whether a particular tool or AI use is appropriate for an assessment, please ask before submitting. I am committed to supporting access while maintaining academic standards."
Education Design Support
Need help? Quick consults, sample Process Logs, viva prompts, moderation packs (internal).