AI Use and Accessibility
AI tools such as Gemini and other generative platforms are increasing the speed and volume of content creation across the university. While these tools can improve efficiency, they do not consistently produce accessible output by default.
This document outlines practical ways to use AI to produce more accessible content from the start, reducing rework and improving overall quality before publication.
Using AI to Create More Accessible Content
If you take one thing from this page, it is this: AI will not produce accessible content unless you explicitly ask it to.
When using AI to generate documents, slides, or web content, include instructions such as:
- “Generate this document using structured headings (H1, H2, H3).”
- “Ensure the content follows a logical reading order.”
- “Provide descriptive alt text for all images.”
- “Ensure tables include clear header rows.”
- “Use plain language and avoid unnecessary jargon.”
- “Write link text that clearly describes the destination.”
- “Structure this content for web use with proper heading hierarchy.”
- “Create slide content with simple layouts and readable text.”
- “Provide a transcript for this video script.”
These prompts improve the starting point, but AI-generated accessibility features (such as alt text or captions) must still be reviewed for accuracy and appropriateness. Any content generated by AI and used in Canvas should also be reviewed and corrected using the Canvas Accessibility Checker.
Using AI with Accessibility in Mind
AI can support accessible content creation when used intentionally:
- Use AI to draft structured content, then review and refine
- Ask AI to simplify complex language for clarity
- Generate draft alt text, then edit for meaning and context
- Use AI to outline documents before building slides or pages
- Treat AI output as a first draft, not a finished product
Use of AI does not transfer accessibility responsibility — content owners remain accountable for ensuring materials are accessible before publication.
Common Accessibility Gaps to Watch For
Even when prompted, AI may still produce issues that require review:
Documents and PDFs
- Missing or low-quality alt text
- Skipped or inconsistent heading structure
- Untagged or poorly tagged PDFs
- Tables without defined headers
Slide Decks
- Missing alt text
- Incorrect reading order
- Low color contrast
- Overcrowded layouts
Web Content
- Improper heading hierarchy
- Unclear link text
- Formatting issues from pasted content
Multimedia
- Inaccurate auto-generated captions
- Missing transcripts
Verification Tools
Consider using the following tools to help guide your initial review, but be sure to verify your work manually to ensure the best experience for all users.
- Documents and PDFs: Use the Microsoft Accessibility Checker for Word and PowerPoint, the Adobe Acrobat Accessibility Checker for PDFs, or Google’s accessibility guide for Docs.
- Slide Decks: Use the Microsoft Accessibility Checker to help identify issues in PowerPoint or Google’s accessibility guide for Slides.
- Website Content: For UIS-managed websites, use the built-in Editorially checker, or use the Siteimprove Accessibility Checker for automated website accessibility testing.
- Canvas Web Content: Use the Canvas Accessibility Checker to ensure content meets structure and contrast requirements.
- Multimedia: Review any captions using the recommended instructions for the software (YouTube, Vimeo, Panopto, etc.) to ensure video content is fully accessible.
Quick Accessibility Review Checklist
Before publishing AI-generated content, confirm:
- Are all images described with meaningful alt text?
- Are headings structured logically and in order?
- Does link text make sense out of context?
- Is color contrast sufficient for readability?
- If exported as a PDF, is the document properly tagged?
- Have captions been reviewed for accuracy?
A brief review can prevent common and avoidable issues.
When to Reach Out for Support
Consult IT Accessibility when:
- Developing custom web applications or tools
- Questions arise regarding remediation approaches or technical standards