AI and the Future of Accessibility
Artificial intelligence is changing how we think about accessibility, not by replacing human expertise, but by significantly expanding what is possible. AI can translate, summarize, describe, caption, and reorganize information faster than ever before. As AI reshapes how content is created and consumed, accessible, human-centered design must remain at the core of that evolution.CMU’s Digital Accessibility Office views AI as a partner in advancing equity—a way to break down barriers, expand participation, and create multiple pathways to information. The power of AI as an assistive technology is often underappreciated and underreported. AI has the potential to provide humans with greater agency and autonomy, which is the fundamental aim of accessibility.
Historically, people with disabilities have driven technological innovation and been among its earliest adopters. AI is no different. People with disabilities are already using AI to simplify tasks and increase productivity.
- Assistive Technology and Independence: Smart assistants, like Amazon’s Alexa or Apple’s Siri, have helped users get online more easily. One example shared by PBS involves a non-verbal child with cerebral palsy using a speaking device to communicate with an Amazon Echo, enabling him to call his grandfather and request a car ride, providing a sense of independence and initiative.
- Information Management: For those with visual and motor disabilities, using AI as a research assistant can consolidate information into a single space, rather than managing multiple browser tabs. Automated tools that manage meeting notes and document summaries can reduce cognitive overload for neurodivergent individuals.
- Creative Expression: Artists who cannot draw conventionally may use voice prompts to have generative AI create their visions, empowering them to indulge in their artistic passion.
- Data Interaction: New AI tools can read, interpret, and manipulate information from data visualizations, allowing users to ask pertinent questions about data points and trends on a chart or graph. It’s interactive data storytelling like we’ve never seen before.
- Employment Impact: The advancement of AI could open up whole new industries to people with disabilities. According to the U.S. Bureau of Labor Statistics, in 2022, approximately 21% of people with a disability in the U.S. were employed, which is the highest rate since tracking began in 2008.
Hot Projects and Future Frontiers
There are many promising accessibility-focused AI initiatives. Below are a few that may pique your interest.
Mobility and Autonomy
- The AI software behind autonomous vehicles is affording many in the disability community (like those with low vision or who are blind) more agency over their transport. Devices like the HOOBOX Wheelie 7 use machine learning to recognize facial expressions, allowing users to command a wheelchair.
Communication and Speech
- Apple’s Personal Voice technology, available in iOS 17, is designed for people with progressive illnesses like ALS, allowing them to record their voice and generate a personalized screen reader that sounds just like them.
- The Voiceitt app uses machine learning to recognize mispronunciations and normalize non-standard speech patterns for those with speech impediments (e.g., due to cerebral palsy or Parkinson's), generating an output of audio or text.
- Augmentative and Alternative Communication (AAC) devices are using AI features to recognize non-standard speech, such as stuttering or speech affected by motor impairments.
Learning and Cognition
- AI can help with executive functioning tasks like organization and time management by breaking down large assignments into smaller steps, creating personalized schedules, and setting reminders.
- Large language models have tremendous potential for student support. They can create accurate, interactive summaries of information, which is a significant advantage for those who struggle with attention and focus.
Ongoing Challenges
AI can be an excellent tool for furthering access, but there are also critical initial usability challenges to consider. New technologies frequently neglect universal design principles. While many AI tools are text-based, making them largely accessible to screen readers, many content creation tools fail to meet comprehensive accessibility standards and have barriers to use with less common assistive technology. More graphical interfaces, such as image generators, are less accessible overall than text-based applications.
While there are strong coding programs powered by AI that can help to generate WCAG-compliant website code, the WebAIM Million Report still reports that 95% of websites have common barriers. Improperly labeled form fields for submitting prompts, lack of keyboard access, and insufficient color contrast for text continue to limit access to information across the web.
AI and Accessibility at CMU
The Digital Accessibility Office encourages the responsible use of AI to reduce barriers, expand participation, and align with CMU’s Digital Accessibility Policy and the Web Content Accessibility Guidelines (WCAG). With human oversight, AI can be useful for learning about accessibility standards and best practices, remediating content, and developing alternative formats to expand access.
In the upcoming articles in this series, we’ll explore some of the tools available with your Andrew ID and how we can practically apply AI solutions to remediate content and create alternative formats.