AI & Accessibility in the Classroom
Let’s face it, everyone is using AI on our campuses - students, instructors, and campus staff. If you’re working on your syllabus or teaching policies for a new class, you may want to think about AI as a tool to help your class, rather than as something suspicious to be banned or ignored. For example, AI tools can assist you in noticing blind spots or problems in the way you select or organize your class materials, in the type of readings and assignments you choose, and in the way you explain things. It can also help you with accessibility tips, to make your class content and assignments more multimodal, and to make things more straightforward for students who might have a varied range of backgrounds, disabilities, or major-specific knowledge. Here are some ways using AI tools can help improve accessibility in your class:
In all cases, treat the output as a draft to review, not a finished product—see the caveats at the end, which matter especially when the work affects students with disabilities.
Syllabus language
If you're trying to make sure your course content covers all its bases when it comes to accessibility, inclusion, and attention to implicit biases, have an AI tool help you as a reading assistant: upload your course syllabus to one of the LLM platforms you're familiar with, and ask it to look for gaps or potential problems. Some tested prompts you can try:
“Please go through my syllabus and let me know if there are any changes I should make to improve accessibility.”
“This is a course syllabus for an undergraduate class with students from various majors. Are there any unexplained assumptions that I should clarify in the description of my class content?”
“Can you spot any biases in my course language or choice of readings?”
Plain language and reading level
Dense or jargon-heavy writing creates a barrier for many students. Multilingual learners, students with certain learning disabilities, and first-generation students may be unfamiliar with academic conventions. AI is good at a plain-language pass that keeps your meaning while lowering the reading burden. Some prompts to try:
“Rewrite this assignment prompt in plain language at about a 9th–10th grade reading level, without losing any of the requirements.”
“Identify any jargon or discipline-specific terms in this syllabus that a non-major might not understand, and suggest a short plain-language gloss for each.”
“Flag any sentences in this document that are long or structurally complex, and suggest clearer versions.”
Assignment and test accessibility
If you're unsure whether your assignments are accessible to your students in terms of format, technical accessibility, and content clarity, you can ask an AI assistant to review what you’ve prepared so far and give you tips on how to edit or enrich your assignments. Here are some prompts we tested for you:
“Copy-paste the text of a specific assignment into your chosen LLM tool, then ask: Are there any changes I can make to the language or format of this assignment to enhance accessibility for students with disabilities?”
“Describe the general character of your class and your learning goals, then ask for domain-specific examples of clear and accessible assignments or tests that respect the general content and goals of your specific class.”
An example if you're teaching a language class: “I'm teaching an advanced beginner Russian language and culture class. Students are expected to learn to discuss their interests, current events, and simple readings. Typical assignments and quizzes include fill-in-the-gap questions, writing short paragraphs, and multiple-choice questions. How can I diversify my assignments and quizzes?”
An example if you're teaching a content class: “I'm teaching a class on Chilean cinema, conducted in English with some Spanish proficiency expected. Typical assignments include multiple-choice questions on historical context, content comprehension questions, and response essays. How can I diversify my assignments to make them more accessible and multimodal?”
Scaffolding and chunking complex work
Large, open-ended assignments can disproportionately disadvantage students who struggle with executive function, including many neurodivergent students. Breaking work into structured, sequenced steps is itself an accessibility move, and AI is well-suited to drafting that structure. Some prompts to try:
“Break this final project into a week-by-week checklist of smaller milestones, with a one-sentence explanation of what each step accomplishes and why it comes when it does.”
“Write a worked example that models the thinking process for this type of problem, step by step.”
“Suggest where students are most likely to get stuck on this assignment, and what scaffolding or check-in I could add at each point.”
Multimodal versions of the same content
A core principle of Universal Design for Learning is offering multiple means of representation, so students can engage with material in the format that works best for them. AI can quickly turn one piece of content into several. Some prompts to try:
“Turn this lecture outline into three formats: a one-paragraph summary, a bulleted study guide, and a short script I could record as an audio overview.”
“Create a concept map (described in text) showing how the key terms in this unit relate to one another.”
“Suggest two or three alternative ways I could present this concept—for example, a diagram, an analogy, and a real-world example.”
Alt text and figure descriptions
Every image, chart, diagram, or slide without a text description is invisible to students using screen readers. Writing alt text by hand is tedious, which is often why it gets skipped—and where AI can help. Some prompts to try:
“Write concise alt text (under about 125 characters) for this image, suitable for a screen reader.”
“This is a complex diagram. Write a longer description that conveys the relationships a sighted student would see, suitable for a figure caption or an appendix.”
Captions, transcripts, and audio
Recorded lectures and videos need accurate captions and transcripts, and some students benefit from an audio version of written material. AI can draft and clean these up—though accuracy-checking is essential here (see caveats). Some prompts to try:
“Here is an auto-generated transcript of my lecture. Clean up the punctuation and obvious transcription errors, and flag anything that looks garbled so I can verify it.”
“Summarize this transcript into timestamped section headings so students can navigate the recording more easily.”
Multilingual support
For courses with international or multilingual students, AI can help you build support without your needing to be multilingual yourself. Some prompts to try:
“Create a bilingual glossary (English–[language]) of the 15 key terms students need for this unit.”
“Point out where in this assignment a multilingual learner might get stuck on phrasing rather than on content, and suggest clearer wording.”
Accessible formatting checks
Beyond wording, the structure of your documents and slides affects screen-reader users and others. AI can coach you through bringing files in line with accessibility standards: true heading styles, meaningful link text, table headers, and not relying on color alone. Some prompts to try:
“Walk me through making this Word document accessible: heading structure, alt text, link text, and table headers.”
“I use color to distinguish deadline types on my syllabus. Suggest a way to convey the same information without relying on color alone.”
An accessibility "stress test" with student personas
A proactive way to find barriers before your students hit them is to have the AI read your materials from several different vantage points. Some prompts to try:
“Read this syllabus as if you were a student who uses a screen reader, then again as a student with dyslexia, and then as a multilingual learner. For each, tell me what would be confusing or hard to access.”
“Review this assignment for barriers a student with limited time, attention, or executive-function support might encounter, and suggest adjustments.”
How to choose the right assistant?
Not all LLM platforms can "read" files. Depending on the task you have in mind, the resources you have, and the tools you're already familiar with, you might prefer to use Microsoft Copilot (UCLA faculty have access to an enterprise version) or Google Gemini. You can use Microsoft Copilot if you want to upload your syllabus as a PDF or Word file, so the AI can directly read it. If you're more at ease with Google Gemini, simply open the file you want to work with and copy-paste the text into Gemini's chat box. When the material is course-related, the enterprise version of Copilot is generally the safer choice for handling your files.
Important caveats when using AI for accessibility
Because these outputs may directly affect students with accommodations:
Review accuracy. AI-generated alt text, captions, and transcripts must be checked; a confident but wrong caption is worse than none at all.
Watch for lost precision. Plain-language rewrites can quietly drop technical accuracy; confirm the meaning survived the simplification.
Supplement, never replace, formal accommodations. These tools do not substitute for working with UCLA's Center for Accessible Education (CAE) or for a student's approved accommodations.
Protect privacy. Do not paste sensitive information (a student's disability status, accommodation letters, or identifying details) into a chatbot.
Standards are the benchmark. "Accessible" is defined by established standards (such as WCAG) and UCLA's accessibility requirements, not by the AI's judgment.