Blog | UsableNet

GAAD and the Impacts of Artificial Intelligence on Accessibility

Written by UsableNet | May 21, 2020 2:47:43 PM

Today is Global Accessibility Awareness Day (GAAD) a day created to bring awareness to the importance of inclusion and accessibility.

GAAD focuses heavily on digital inclusion, which means ensuring that your entire online presence (mobile apps included) is accessible for everyone. Due to COVID-19 there are a large number of free online virtual events. 

As technology has advanced, it’s opened up new potential opportunities and challenges alike for businesses. Artificial intelligence and machine learning are great examples of this.

For this special post on Global Accessibility Awareness Day, we interviewed Joseph DiNero about his experience with AI and Assistive technology as a person with a visual disability. Joe is the Head User Tester for UsableNet and a teacher of Assistive Technology at Helen Keller Institute.

We've shared the Zoom audio from this interview here with captions from Youtube, listen or download a copy of transcript.

In our interview and in today's post, we take a critical look at how machine learning and artificial intelligence (AI) are impacting accessibility today and what it could mean for accessibility in the future.

The Emergence of Mainstream AI Devices 

People who are using assistive technology like screen readers or automated voice assistants have more tools available to them than ever before.

The number of smart assistants alone are outstanding and offer incredible functionality. Siri, Amazon Echo, Cortana and are all easy-to-use and highly accessible. You can ask your smart assistant to call or text someone, to look up a local business, help you navigate back home, or even to access digital content like music.

Devices like Amazon’s Echo even allow you to use voice control to manage smart home features, like turning off lights or ensuring that smart locks are in fact locked.

This is a huge advantage, and brands must remember to incorporate voice search optimization into their marketing and SEO strategies.

The Limitations of AI for People with Disabilities

While the emergence of new artificial intelligence and machine learning tools are an enormous asset to many with disabilities, it’s essential to acknowledge that they do have their limitations.

You could theoretically use Siri, for example, to check your bank account, but only if it’s able to gain access to the bank information through the mobile app.

As Joe explained, that’s where the problem comes into play; he can open his bank’s mobile app with Siri, but he needs to revert back to voiceover and screen readers to actually log in, check their balance, deposit checks, and make deposits. Since depositing a check, for example, requires you to be able to read a paper check to enter in the amount needed to deposit, there can be accessibility issues.

As useful as artificial is, it’s essential that businesses remember that the onus is still on them to ensure that their websites and mobile apps are accessible and can be used with other types of assistive technology. This means that you must be following standard protocol, including guidelines from the WCAG. The app itself needs to be accessible.

Which WebSite and Apps Present the Most Challenges

When Joe consumes content online, he is fully reliant on assistive technology. In his experience, there are some types of websites that are generally easier for him to use than others. For now, most of these challenges are not easily identified or corrected with AI technology. 

E-commerce sites typically have functionality that could cause issues for users relying on assistive technology. This is largely because there are a lot of interactions involved to complete a purchase, including searching, reviewing, and selecting the right product, adding it to your cart, and completing the checkout process. Artificial Intelligence cannot fix these issues.

Joe also noted, however, that even sites with general content are sometimes not structured well, even if it’s full of plain text. This makes navigation an enormous challenge, especially if there isn't effective navigation or clear headings. A layout that may seem intuitive to someone with full sight can be chaotic and impossible to understand for those using screen readers if not coded correctly; for this reason, it’s always best to hire those with disabilities to test and audit your site’s accessibility. 

Automated AI for Web Accessibility: Do They Work?

There are some solutions in the web accessibility market that claim to automatically remediate all accessibility issues immediately with no need for the web site owner to do anything. They’re marketed to businesses and web developers who are trying to make their sites accessible as quickly as possible and have little time to validate the claims.

Unfortunately, these solutions typically don’t live up to their claims and their widget features are being added to lawsuits claiming sites are inaccessible.

They fail to address most major accessibility issues, such as keyboard navigation, even if they claim to be able to use machine learning for this purpose. DiNero explains, “their functionality is restricted, which can actually limit access to the site for me as I use assistive technology that does not benefit from the accessibility features the widget offers. Only an actual user testing the site with assistive technology could catch the important issues to a screen reader user like myself and offer suggestions for how to improve it.”

For the Virtual Conference, "A Future Date," we talked with Joe and Tanner Gers from the American Foundation for Blind about how to engage people from the disability community to test your website and apps. If you're interested in learning more on this topic, Watch the presentation here or download the slides. 

Conclusion

Artificial intelligence and machine learning are rapidly developing, offering new potential opportunities for users and businesses alike. It offers up huge benefits in the broader accessibility space but unfortunately AI and machine learning that currently exists to automate accessibility on websites is relatively limited. In the website accessibility market, it is being over-hyped as a solution for all.

In the future, we hope that AI and machine learning will be able to improve and automate accessibility for more website, but in the meantime, the best solution is to work with people from the disability community to ensure that your site is fully accessible for your entire audience.

Interested in learning more about how artificial intelligence and machine learning is impacting the world of accessibility today? Listen to this interview with Joe and Jason here or download the transcript here.