AI, WCAG 3.0, and Cognitive Accessibility: What I Learned at AccessU

Did you know that if you write something in Spanish online but don’t use the proper language tags in HTML, a screen reader will say it with an American accent? The result sounds something like Peggy Hill from King of the Hill. It's funny … but probably not the experience you were hoping to create on your site.

I didn't know that until I went to AccessU, where I saw and heard it happen live. John Slatin AccessU is a conference focused on digital accessibility, put on by Knowbility and held every year in Austin, Texas. It's aimed at designers, developers, content creators, program managers, and accessibility advocates working to make digital experiences more accessible.

This was my first year attending, and I loved it. I've had a few lackluster experiences at other conferences lately, but AccessU felt different. The quality of the sessions and the strong sense of community made it feel like it was worth the time.

After three days of sessions, three topics really stood out to me:

• AI and digital accessibility
• The future of digital accessibility standards (WCAG 3.0)
• The accessibility of digital documents (PDFs, presentations, and trainings)

My main takeaway from the week: as long as we continue to center humans in our work, the future of digital accessibility is one of hope. And that future will make the work lighter for everyone involved.

How will AI change accessibility testing? 

At AccessU, I attended a session titled "Where AI Testing Ends and Human Responsibility Begins," which outlined what parts of the accessibility testing process AI does well and what it struggles with. 

AI is limited in what it can properly audit. It doesn't do a great job of judging human context, like how a page or feature fits into a broader user journey. It struggles with dynamic features that require someone to trigger an interaction and observe the change, like modal dialogues, dynamically loaded content, and form error validation. 

And AI can't meaningfully emulate testing with assistive technologies like a screen reader.

In another session on screen reader testing, accessibility specialist Deneb Pulsipher reminded us that users who are blind navigate websites in ways far more nuanced than most sighted developers assume. They don't read top to bottom. They scan links and regions first to quickly build a mental map of the page, navigating nonlinearly, similar to how sighted users scan. He suggested trying to use a screen reader without looking at your screen to get a better sense of the experience. AI can't replicate that kind of contextual, experiential testing.

The range of how AI can support accessibility work is broad, from asking an LLM questions about your code to building automated pipelines that check for accessibility errors during development. As AI advances and evolves, how it affects digital accessibility work changes every day. 

Luckily, most of the folks at AccessU already had a nuanced understanding of automatic accessibility checkers and the limitations of those platforms. Automatic checkers can typically find only ~30-60% of accessibility issues on a site. While AI will make automated checkers faster and better, we still need humans involved to find the other half of errors. 

What is the future of WCAG?

Shawn Lawton Henry of the W3C gave a keynote at AccessU focused on what's coming with WCAG 3.0, the next major version of the W3C Accessibility Guidelines. The main takeaway: WCAG 3.0 is still years away from official release.

Even though we still have years to wait, the keynote did get me excited about when 3.0 eventually lands. 

The first improvement is kind of a meta-improvement. The WCAG 3.0 guidelines themselves will be written using plain language. Gone will be the days of explaining what plain language is, using documentation that is clearly not written in plain language. 

WCAG 3.0 introduces something called "assertions" to say what activities you did to meet guidelines that don’t fit neatly into a binary pass/fail. So, for example, to demonstrate your site’s content is readable and understandable, you can use an assertion to say you ran user accessibility testing or implemented training with your content writers.

The new spec also rethinks how compliance levels work. Right now, most laws point to Level AA, which means Level AAA criteria have gone largely ignored. WCAG 3.0 is being designed to better motivate organizations to improve their sites' accessibility beyond the baseline. We all know that accessibility work is a journey, and it sounds like WCAG 3.0 will really live that idea.

Lastly, WCAG 3.0 will have a larger focus on cognitive disabilities. That focus felt especially timely given another session I went to called "Designing for Optimal Cognitive Function." One of the presenters, Dr. Anne Forrest, shared her experience as a person with a traumatic brain injury. She described having an instantaneous negative reaction to a website that used black text on a yellow background. For her, seeing yellow or red colors in a design communicates a feeling of caution or danger. Her experience was documented in the book "Maximum Accessibility" by John M. Slatin and Sharron Rush (2002):

“Even after changing the style sheet to display white text on a blue background, this informant explained that for her, the multiple colors used to differentiate various sections of the form were simply additional information that she had to take into account. She reported that the "background" colors never receded into the background but remained in her consciousness and demanded attention each time she looked at the screen. The effect was to siphon off energy that she would rather have used to read the items on the form and to make the decisions that judging requires.”

The range of what "accessible" means for people with cognitive disabilities is vast. Color, layout, information density, motion, language complexity: all of it varies enormously from person to person. I'm hopeful that WCAG 3.0 will give designers and developers clearer guidance for navigating this landscape.

Document Accessibility (PDFs, Presentations, and Training)

The conference inspired me to create a personal goal for the year: earn IAAP's Accessible Document Specialist (ADS) certification. While the focus of my career has been web technologies, I produce a lot of presentation decks and training materials in my work these days. So when I was picking sessions at AccessU, I looked  for sessions about PDF remediation, accessible presentations, and training formats.

The first thing I learned was that the accessibility community does not like PDFs. Many folks started their careers in digital accessibility by doing PDF remediation, and the stories of long days spent remediating documents, only to have them replaced by a new version with a fresh crop of accessibility errors, were enough to make anyone sympathize. But PDFs aren't going anywhere, and understanding how to make them (and presentations and training sessions) accessible will be an increasingly important part of my work at Capellic.

I picked up a lot of practical strategies from these sessions that I'm planning to write about separately as I work toward that certification. For now, I'll just say: if you've ever agonized over whether your slide deck or PDF is actually usable by everyone, you're not alone, and there are clear paths forward.

Maybe join me in Austin next year?

Everyone at AccessU seemed to share a goal of making our world more accessible. It was validating to talk to people who have faced similar digital accessibility barriers in their organizations and codebases.

I'm walking away with new skills and new ways of thinking, both to help me in my day-to-day work at Capellic and to prepare me for my next IAAP accessibility certification.

During one of the keynotes, civil rights attorney Eve Hill said we were all collectively pushing back against "thoughtlessness elevated into systems." I look forward to continuing that fight at Capellic, with our clients, and maybe with you in Austin next year.