Accessibility means ensuring that all learners, including students with disabilities and other barriers to learning, can access and engage with all course materials, activities, and assessments. Employing the Universal Design for Learning (UDL) Framework can help make your course more accessible. A teaching approach and course design framework, UDL works to accommodate the needs and abilities of all learners and can help eliminate unnecessary hurdles in the learning process. 

As you work to make your course more accessible, it’s important to ensure that your course materials can be used with a robust set of assistive technologies to ensure that you are creating an equitable space for all learners. generative AI tools can help you with this process. However, note that your course can also be made fully accessible with minimal use of generative AI. For more information about how to do this, review the CTI Accessibility Guide.

If you choose to use generative AI as a tool, it’s important to consider the following:

Accessibility and AI Literacy 

The rise of generative AI technologies introduces a new facet to making courses accessible. In fact, when thoughtfully employed and driven by a knowledgeable human, generative AI can function as an accessibility tool and resource.

However, as always, the need for generative AI literacy, particularly as to how it applies to accessibility, is essential. While ChatGPT and other AI assistants can help make course materials more accessible, they cannot replace the essential role of human judgment in the process. When using generative AI, a human’s judgment in deciding what to do to make a material accessible—and to review the generative AI-outputted material with knowledge of exactly what makes a resource accessible—is needed at nearly every step of the process. 

Pitfalls with Relying Fully on Generative AI

Using generative AI to generate large chunks of information has a number of pitfalls to be aware of. One of the most serious concerns is the potential of generative AI-produced content to include inaccurate representations of individuals with disabilities or individuals who are neurodivergent.

There are also implications for instructors using generative AI to help make their courses more accessible. generative AI tools are not yet reliable at making a material fully accessible on their own – they require a human to guide the process. Asking ChatGPT or another AI assistant to “make this document accessible” does not guarantee an accessible result. Check the output against accessibility requirements and test it with relevant assistive technologies.

Because of this, if you choose to use generative AI for the purpose of making your course more accessible, it is best to limit your generative AI use to prompts for individual steps of the process, and not ask it to do the work for you.

When Incorporating Generative AI Tools into the Curriculum

When learners are required to use generative AI for an assignment, the tool itself should be keyboard-navigable and accessible to assistive technologies, such as screen readers.

To determine whether a generative AI tool is accessible, there are a few things you can do. 

• Conduct your own research (e.g., search for a Voluntary Product Accessibility Template, or VPAT, for the product). A VPAT is a document that vendors can fill out that outlines clearly where their product meets or misses the accessibility guidelines. Note that interpreting this document may require some knowledge of accessibility standards.  If you have any questions about a VPAT or whether a vendor tool meets these standards, you can request support from the Cornell IT Web Accessibility Team via the Cornell IT Service Desk.
• Request the tool be reviewed via Cornell IT’s Vended Product Review Process. If you plan to use an AI tool that has not been reviewed for accessibility, consult the Cornell IT Web Accessibility Team about the Vended Product Review Process. They will review the tool for you and give you expert advice on whether it meets accessibility guidelines.