AI Accelerates Development But Accessibility Testing Lags
AI has sped up design and development, but quality assurance testing hasn't kept pace, potentially making the web less accessible. Researchers are looking for ways to embed human judgment into AI-enabled workflows.
Original reporting · smashingmagazine.com
Original: “AI Sped Up The Build: How Can Accessibility Testing Keep Up?”
Read “AI Sped Up The Build: How Can Accessibility Testing Keep Up?” at Smashing Magazine →
How this was written. Awesome B2B Design wrote this article with a language model from Smashing Magazine’s reporting (model: @cf/meta/llama-3.3-70b-instruct-fp8-fast); it describes that reporting rather than adding to it. The original, linked above, is the full story.
Key points from Smashing Magazine’s report
- AI has accelerated design and development, but QA and testing have not kept up, potentially making the web less accessible.
- Teams that have already shifted left on accessibility are off to a good start, but need to re-evaluate their workflow to keep up with AI's pace.
- Building human judgment into the workflow is crucial to ensuring accessibility keeps pace with AI-accelerated development.
- Mature accessibility practices, such as training and communication, are key differentiators in improving accessibility metrics.
- Teams can integrate accessibility checks and judgment calls into their existing workflow to avoid shipping accessibility barriers or spending time fixing them later.
AI has accelerated the design and development process, but quality assurance testing has not kept up, according to Smashing Magazine. This discrepancy may lead to a less accessible web. Researchers are exploring ways to integrate human judgment into AI-enabled workflows to address this issue.
The 2026 State of Digital Accessibility Report by Level Access found that AI has helped teams speed up planning, design, and development, but QA and testing have lagged behind. Only 30% of professionals surveyed said AI had sped up the testing stage, compared to 58% for planning and 61% for design.
The report's findings suggest a two-speed workflow, where design, planning, and development are accelerated by AI, while checking and validation are still done at a slower pace. This can lead to teams rushing through the build process only to be slowed down when checking for accessibility.
Accessibility has historically been treated as a QA bottleneck, and AI is pushing more work into it, faster. Teams that have already shifted left on accessibility are off to a good start, but even proactive teams need to re-evaluate their workflow to keep up with AI's pace.
The web is currently moving in the wrong direction, with automatically detected errors per home page rising 10.1% in a year, according to WebAIM's latest scan. AI-assisted coding is cited as a possible contributing factor. To address this, researchers are looking for ways to build human judgment into the workflow as AI continues to accelerate the pace.
Researchers found that AI adoption made a small difference in improving accessibility metrics. Instead, the maturity of surrounding accessibility practices, such as training, communication, and a culture of accessibility, was a key differentiator. Professionals in mature accessibility programs were more likely to say AI helps them prioritize issues and tackle accessibility more proactively.
To avoid shipping accessibility barriers or spending time fixing them later, teams can build the right accessibility checks and judgment calls into their workflow. This can be done by involving people at three points: finding issues, fixing them, and documenting the evidence. Each point can be integrated into existing steps, such as handoffs, pull requests, or reviews, to keep accessibility in pace with the work in motion.
Teams can start by putting judgment checkpoints close to the creation work, such as at the handoff or pull request stage. Automated checks can be run in the CI/CD pipeline, and human judgment can be applied to review changes and assess usability. By giving every check an owner, teams can ensure that accessibility is integrated into the workflow and that human judgment is applied at every stage.
Questions and answers
What is the current state of accessibility testing in AI-accelerated development?
Accessibility testing has not kept pace with AI-accelerated design and development, potentially making the web less accessible.
How can teams build human judgment into their AI-enabled workflows?
Teams can involve people at three points: finding issues, fixing them, and documenting the evidence, and integrate these points into existing steps such as handoffs, pull requests, or reviews.
What is the key differentiator in improving accessibility metrics?
The maturity of surrounding accessibility practices, such as training, communication, and a culture of accessibility, is a key differentiator in improving accessibility metrics, not AI adoption itself.


