Skip to main content

AI and digital accessibility: Revolutionising inclusion or creating new barriers?

Posted by Lucy Collins on May 2, 2024 4:37 PM

Artificial intelligence (AI) is having a bit of a moment. Every other Linkedin post, email campaign and conference invitation seems to focus on this hot topic. But what does it mean for digital accessibility?

We talked recently about how we at Web Usability hope to navigate the AI revolution and expressed both our hopes and concerns for the AI dominated future that many are predicting.

Ahead of Global Accessibility Awareness Day, we also wanted to reflect on AI’s impact on digital accessibility…

  • Will AI improve access to and inclusivity of the digital world?
  • Will it make it easier for us to build and maintain accessible online services?
  • What potential issues should we be aware of?

How AI will help people using your website

One of the very exciting things about AI is how it can personalise the experience that people have online.

Human beings are unique and so are the disabilities we experience. In the UK, the Department of Work and Pensions estimate there is somewhere in the region of 16 million people with a disability; that’s nearly a quarter of the UK population.

Each one of these people and their disability is unique meaning they also have a unique set of needs that need to be met.

The great advantage of AI is that it can adapt online interfaces and content delivery to meet these unique needs.

Case study: How AI can be my eyes – Adi Latif, Blind Accessibility Consultant

“AI really excites and more so at the moment because its tangible, it’s something we already have and can use now.

The Be My Eyes app, for example, is great. I can be in a restaurant and take a picture of the menu and it can explain it to me. And it doesn’t just give me all the text but it explains it to me in really intelligent ways. Instead of reading everything out to me, it will tell me the various categories on the menu and here’s a few items from each of the category. Then I can ask it questions like “oh tell me the vegetarian options” and then it’ll list vegetarian options.

With the same app, I can take a picture of the restaurant [to get a sense of what it is like]. Today when I was in a cafe, I was sitting on a high bar stool and I was just curious if there was other chairs or tables of a regular height… so I took a few pictures and then it described the layout of the restaurant to me, everything from the paintings on the walls to the way the tiles were and that there were tables and chairs.

I was able to find out all that information using AI.”

Some other examples of how AI is helping people with disabilities include:

  • Microsoft Seeing AI: A mobile app that uses AI to describe nearby people, text, and objects. It can read printed and handwritten text as well as describing a person’s surroundings. A great tool for people with visual impairments to get a better sense of the world around them.
  • Predictable: Another app, this time an AI-enhanced communication tool that can help those who have lost the ability to speak. By learning how a user communicates, it can predict what they might want to say next, making communication faster and more efficient.
  • AVA: A live captioning tool that allows deaf and hard of hearing people to engage in conversations, both in-person and online, in real-time.

How AI will help you build and maintain websites

All these tools and technologies will likely be hugely empowering for people with disabilities and the AI enhancements should only make them more effective. However, we cannot rely on them.

We (and I mean all researchers, designers, developers and content editors) still have an obligation to deliver websites and digital services that are as accessible and inclusive as possible.

The good news is, AI has also been making waves here and there are a number of tools that look like they can help ensure accessibility is built into the work that we do.

  • Automated accessibility checkers: a range of these tools are already available on the market and a number are now making the most of AI enhancements to better cope with some of the subjectivity that is required when interpreting the accessibility guidelines. Deque’s Axe tool is one such example that we use as part of our more comprehensive manual testing.
  • Content optimisation tools: AI can also help us to develop content in an accessible way. It can improve how we write text so it works for people with neurodiversity or craft alt text to ensure blind people can still access visual content.
  • Predictive interfaces: AI’s ability to process vast amounts of data offers great potential for predicting user behaviour. This in turn may reduce cognitive load and physical interaction required to engage with a digital service, by offering time-saving tools like autocomplete and customised interfaces.
  • Natural language processing (NLP): Being able to talk to machines like they are human, rather than having to know complex code or commands will also improve both how we build and use digital services. Smart assistants, like Alexa and Google Echo are classic examples of how simple commands can be used very effectively.
  • Developer tools: For the technical folks among us, AI is also being used to during the coding process to ensure digital services are built in an accessible way from the ground up. GitHub’s CoPilot is one of the most well known.

AI and accessibility: issues we should not overlook

But before we get too carried away, I feel no AI-focused blog article is complete without taking time to reflect on the limitations or challenges this technology presents.

In the world of AI and accessibility, the key ones for me are…

  • AI does not replace human beings: Human beings, as well as being unique, are unpredictable. When it comes to considering the specific needs of people with disabilities, nothing beats talking to those who have the lived experience. AI may be able to simulate what we want and how we’ll behave but it is not the real thing.
  • Only as good as the data it is trained on: It is a well-known reality of AI that the beast will feed on what you give it. If the data input is incomplete, biased or inaccurate then the resulting outputs will also be skewed. This could result in inaccurate alt text or poorly written content that may be, at best, misleading and at worst, damaging.
  • Creates lazy design and development: With all these tools and technologies at our disposal, there is a danger of relying too much on AI to ‘fix’ accessibility problems. Understanding inclusive design principles from the start is crucial, we cannot rely on AI to retrofit accessibility.

Accessibility and AI make a good pair. Many of the developments offer empowering new ways for people with disabilities to navigate the online world. New tools and technologies will also support those building digital services to do so in a more inclusive way. However, we should not hand over all control and remember that people remain at the heart of good online design.

Subscribe to our newsletter

Sign up to receive regular updates straight to your inbox.