AI and Accessibility

Starting the process of creating accessible digital materials, can be overwhelming, as there are many laws to learn and guidelines to implement. WCAG 2.1¹ alone has 78 success criteria, with multiple techniques for applying each one. If you are new to the process of creating or making your course materials more accessible, this can feel like a daunting challenge. AI-powered tools can help make this process feel less overwhelming.

• For example, when you have an accessibility question, instead of trying to sift through many pages of legal and technical documents or pages of Google results, you can use an AI-powered tool to refine your search and go directly to the guidelines you need.
• Example prompt: Using a tool such as Microsoft Copilot, type, “Which WCAG 2.1 guidelines relate to links?” 

You can then use this information to make your learning process more focused, only researching the guidelines for specific topics you currently need. This same technique can be applied to researching laws and other guidelines/frameworks.

This section looks at how you can use AI-powered tools when remediating digital materials, speeding up the remediation process and saving you time. Each of these strategies can be done with the Cornell-supported tool Microsoft Copilot, but you can use these prompts or variations of them with other tools as well.

Here are some examples of tasks you may need to perform to make course materials more accessible, and how generative AI can help with these tasks.

• Writing long descriptions for complex images:
  • Complex images, such as charts, graphs, infographics, and maps, require long text descriptions to make them accessible. Some courses have dozens of complex images, and it can take many hours to accurately describe each one. AI-powered tools can generate a draft description for an image that you can then edit for accuracy, significantly cutting down the time needed for this task.
  • Example prompt: Upload an image and type, “Describe this image in 4-5 sentences.”
• Simplifying language:
  • Text is most accessible when it is clear, concise, unembellished, and describes and defines all acronyms and jargon. This is especially true for your syllabus, assignment instructions, and guidelines: Ensuring these documents are worded clearly and comprehensible is vital to ensuring learners understand and can fully participate in the course. AI-powered tools can take a sample of text and simplify the language.
  • Example prompt: Copy/paste a paragraph and type, “Simplify the language in this paragraph, spelling out all acronyms.”
• Creating a scaffolded course schedule for a project:
  • For multifaceted assignments, it can be difficult for some learners to start early and give themselves enough time to complete each stage of their work. You can help by scaffolding the assignment into many small assignments due throughout the semester and by using an AI-powered tool to quickly generate a schedule for students to follow.
  • Example prompt: Type the following prompt, adding your own deadline: “For a paper consisting of a rough draft due X, a bibliography due on Y, a second draft due on Z, and a final draft due on AA, create a schedule.”
• Generating key points or a summary from a text:
  • Study guides and other aids are useful to all learners, especially for final examinations. You can generate a list of focal points or summaries of key readings from the course and create a helpful reading guide for learners to consult while studying.
  • Example prompt: Copy/paste a selection from a reading and type: “Summarize this text in 2 paragraphs or less.”
• Creating a transcript out of a captions file:
  • Videos often need both captions and a transcript to be accessible. For reasons why both are needed, review Accessible Media. Remember that not all video platforms automatically convert a captions file into a transcript. However, you can use an AI-powered tool to remove the timestamps from a caption file and convert it into a nicely formatted transcript.
  • Example prompt: Copy/paste a selection from a caption file and type, “Extract only the text from this caption file.”
• Getting ideas on how to fix inaccessible code/generate accessible HTML code:
  • This guide focuses on strategies for files and media, but there are times when faculty and students find themselves having to do some programming, for example, those in Computer Science. As you research more about software and web design, you may find opportunities and reasons to learn more about accessible code. Some accessibility tools will alert you to issues in your code, but it can be challenging to find solutions that still fit your design. You can use AI to generate starting code and modify it to fit your aesthetic, or you can prompt AI to fix code you already know is inaccessible.
  • Example prompt: Copy/paste your code and type, “Make this code accessible.”

The next set of ideas and tools will look at how AI can help you remove barriers in the classroom through your teaching practices, which has the benefit of both helping your students and saving yourself time. Many of these AI tools and enhancements address common accommodations, such as notetaking, and can apply to large and small classes.

Here are a few examples of where students may face barriers to their learning, and how generative AI tools can help you remove these barriers. 

• Supporting learners having trouble hearing or understanding your lectures:
  • Automated live transcription and captioning: AI-powered tools can automatically transcribe spoken content in lectures and meetings, as well as generate captions for videos. Using these tools, you can create a transcript to share with students before the lecture to better assist them with following along with the content and discussion.
    • Tools: Cornell Zoom, Microsoft Powerpoint, Otter.ai*², Rev.ai*
  • Automated note-taking: AI-driven speech recognition software can act as a note-taker during lectures or meetings. For those who cannot participate in traditional note-taking methods, this can be particularly beneficial in making sure they have accurate notes for the lecture.
    • Tools: Google Docs, OneNote, Speechnotes*, Dragon NaturallySpeaking*, Evernote*
  • Language translation: AI translation tools enable you to translate course materials, communications, and instructions into various languages. This can help in fostering inclusivity among students with diverse linguistic backgrounds.
    • Tools: Microsoft Copilot, OpenAI ChatGPT*, Google Translate*, DeepL*
  • Text summarizing: AI algorithms can summarize lengthy documents or articles, extracting key information and making it easier for individuals to digest large amounts of text. Using these summaries as an outline to share with students who may struggle with taking notes can offer assistance in making sure they have written access to key points of the lecture.
    • Tools: Microsoft Copilot, ChatGPT*
• Checking the accessibility of materials as you create them: There are a number of tools that can help you catch multiple accessibility issues at once. These tools can check an item that is already in Canvas or a document stored elsewhere, and help to ensure that these materials are accessible for students before sharing. This helps to remove some of the barriers students may face when trying to access it.
  • Tools: Canvas Accessibility Checker & Ally
• Making your workflow more accessible: AI-driven virtual assistants can handle routine administrative tasks such as scheduling meetings, organizing files, and sending reminders, often through simple verbal commands. Using AI to help create these tasks can help with alleviating cognitive load and information processing barriers, enabling you to focus more on instructional activities.
  • Tools: Google Assistant, Microsoft Cortana
• Identifying learners who are falling behind: AI analytics tools can analyze student performance data to identify patterns and trends. Reviewing this information enables you to provide targeted support and interventions for students who may be struggling academically or emotionally.
  • Cornell tools: Canvas New Analytics
• Providing an equitable experience for those who can’t participate: AI-driven VR simulations can create immersive learning experiences that engage students and provide opportunities for hands-on learning, which can be particularly beneficial for students with physical disabilities or those who require alternative learning environments. Also, AI large language models such as Microsoft Copilot can provide alternative activities or modifications allowing for all to participate.
  • Meet with an instructional designer in our Creative Technology Lab’s Immersive Teaching and Learning Space
  • Example prompt: “List 5 ways I can modify this activity to work for students with x barriers and still be equitable.” Or “List 5 alternative activities that still accomplish x learning outcome”
• Searching information quickly: AI algorithms can search and retrieve relevant information from vast databases or the internet, helping users find answers to their questions quickly.
  • Example activity: Have students search for items specifically using a generative AI tool, then ask them to perform the same search via Google, and compare the results. Integrating this variety in search results can show students both the powers and drawbacks of using AI.
  • Tool: Microsoft Copilot
• Creating content and lesson planning: AI-powered content generation tools can assist you in creating instructional materials, lesson plans, and interactive multimedia content, while also incorporating principles of Universal Design for Learning (UDL) into the materials.
  • Tool: Microsoft Copilot
  • Example prompts:
    • “List 10 ways I can make this lesson more UDL-friendly.”
      • Follow-up prompt: “Identify which of these ideas are the easiest to implement and impact the most students.”
    • “What are 10 active learning ideas to incorporate on the topic of x that are inclusive of various disabilities?”
    • “Give me 10 ways to engage students on x topic that still accommodate different learning styles and physical abilities.”

The following strategies and prompts offer ways learners can use AI tools as a study aid, or to quickly make their own learning experiences more accessible, which can empower them to take control of their own learning.

• Generating descriptions of images that don’t have alt text: Some images used may not have alternative text associated with them. Depending on where the images live –in a lecture, a PDF, or on a website – you and your students can utilize software to assist in creating alternative text to help better describe the image.
  • Example prompt: Upload an image and type, “create alternative text for this image.”
• Creating a study/homework schedule: Students can leverage AI’s ability to organize data by entering due dates, social calendars, and other important information, and asking an AI tool to create a schedule to help organize all of this information.
  • Example prompt: “I have a rough draft due next week for Literature with the final draft due at the end of the month for a 10-page paper. I also have a Chemistry test next week on Tuesday and a two-day observation for my Education class the following week. Please make me a schedule to help organize this.” 
• Summarizing a reading to understand the main point better: AI can summarize lengthy documents or articles, extracting key information and making it easier to digest large amounts of text.
  • Example prompt: “Can you summarize the following text for me…” 
  • Follow-up prompt: “Can you create an outline from this summary for note-taking?”
• Translating difficult language/text into easy-to-understand language: Drawing upon AI’s vast databases, complex or unfamiliar language/texts can be broken down into easier-to-understand language. AI can take confusing or complex thoughts and help make them into something more familiar to the user whether that be a different language or more commonly used words.
  • Example prompts:
    • “Can you translate the following passage into French”
    • “Write this article in more standard English”