This interactive workbook contains multiple workflow tabs. Use the tab navigation above to switch between sections, or use the sidebar navigation to jump to specific content.

Inclusive and Accessible Learning and Teaching with Generative AI

This workbook is designed for HE teaching staff who want to use Generative AI to create, adapt, and review teaching materials in ways that are inclusive, accessible, and practical across disciplines.

It introduces the UDL framework, explores what an inclusive university environment looks like for students, and shows how GenAI can help staff build better materials from the outset rather than retrofit them later.

You can work directly in Copilot or another approved AI tool while using the workbook.

Use GenAI to remove barriers, not add them.

The key question for learning and teaching is not whether AI can generate content quickly. It is whether that content helps more students participate, understand, and demonstrate learning in equitable ways.

Academic judgement remains essential. GenAI can suggest, scaffold, summarise, and transform content, but staff remain responsible for accuracy, ethics, accessibility, and pedagogic fit.

Activity: Basic UDL-Focused Prompting

Design for the Margins

A strong prompt tells the AI what you are trying to teach, who the learners are, what barriers you want to reduce, and what form the output should take.

This table describes practical prompt 'ingredients' for GenAI to create inclusive and accessible teaching materials.
Prompt ingredient What to specify Example Useful for
Audience Who the material is for, including level and likely prior knowledge "First-year students encountering theory-heavy reading for the first time" Making explanations and examples more appropriately pitched
Teaching purpose What the material is meant to do in the learning sequence "This is a pre-seminar guide intended to prepare students for discussion" Aligning outputs with learning design rather than generic content generation
Barriers to reduce Which access, comprehension, or participation barriers to address "Reduce jargon load, add structure, and provide an alternative to a dense PDF" Embedding inclusion and accessibility from the first draft
Output format What form the response should take "Produce slide text, speaker notes, alt text, and a 150-word student summary" Generating reusable materials across different teaching formats
Quality criteria State expectations such as readability, tone, accessibility, or alignment with policy "Use plain language, clear headings, and include a short glossary for unfamiliar terms" Improving consistency and accessibility in AI outputs
The UDL Prompt Pattern

Use this structure to ensure your AI outputs support inclusive design:

Context: I'm teaching [subject] to [student level] students with diverse learning needs.
UDL Goal: Create content that provides [specific UDL principle - engagement/representation/expression]
Content: [Your original content]
Request: Please [specific transformation needed]
Accessibility Note: Ensure the output is accessible to students who may [specific considerations]

Example:

Context: I'm teaching cellular biology to first-year undergraduates with diverse learning needs.
UDL Goal: Create content that provides multiple means of representation
Content: "Mitosis involves chromosome replication and nuclear division"
Request: Please create 3 alternative explanations: one using everyday analogies, one as a step-by-step process, and one focusing on the biological significance
Accessibility Note: Ensure explanations avoid metaphors that require specific cultural knowledge

⚠️ Important Limitations to Consider

  • They may reproduce biases from their training data, including racial, gender, cultural, and ableist biases
  • They cannot know your students : you provide that contextual knowledge
  • They may produce culturally homogenised content that centres Western norms by default
  • They may hallucinate facts, names, or references, especially in less-documented areas
  • They raise data privacy concerns : never input identifiable student data or confidential university content