A new survey from University of Phoenix and The Harris Poll finds that 89% of recent working learners see ways AI could improve accessibility in their workflows, but 45% say workplace AI policies either lack clear accessibility guidance or leave them uncertain about what is covered. The survey of 1,019 U.S. employed adults who completed a training or school course in the past 12 months was released ahead of the 36th anniversary of the Americans with Disabilities Act (ADA) on July 26.
The findings focus on accessibility in digital environments-ensuring content, tools, and AI outputs are usable by people with different abilities through inclusive design, assistive technology, or conformance with standards like the Web Content Accessibility Guidelines (WCAG). The survey highlights the growing intersection of AI and accessibility in educational environments, a topic explored in resources like AI for Education.
"The reality is that accessibility benefits everyone," said Kelly Hermann, vice president of accessibility and student affairs at University of Phoenix. "If accessibility is built in from the beginning, organizations are more likely to create AI-enabled environments that are universally usable. Clearer content, better summaries, accurate captions, and multiple formats can help workers and learners with disabilities, but they also help busy adults, multilingual learners, mobile users, and anyone trying to absorb information quickly."
What workers see in AI accessibility tools
The survey identified several areas where working learners believe AI and accessibility tools can make a difference:
- 38% said creating accessible documents, presentations, websites, or learning materials
- 33% pointed to presenting information in different formats such as plain language, audio, summaries, or translations
- 30% cited training employees or learners on accessibility practices
Among those already using AI at work, 60% said it has improved their knowledge of accessibility standards and guidelines. Yet only 27% of respondents said the AI tools available through their workplace or professional learning environment support people with disabilities very well.
Policy gaps and the need for human oversight
The survey exposes a clear gap in organizational guidance: 45% of recent working learners said accessibility is either absent from, unclear in, or not explicitly covered by their workplace AI policies. This lack of clarity arrives as AI tools are increasingly used to draft documents, generate captions and transcripts, and create image descriptions-tasks where poor implementation can create barriers rather than remove them.
Human judgment remains a key part of responsible AI use for many respondents. 36% said human review for important decisions or high-impact work should be part of responsible AI practices at work or school. Hermann reinforced that point: "Responsible AI is not only about productivity. It is about whether the technology works for the people who need to use it. AI can help create more accessible materials and more flexible ways to engage with information, but it still requires clear policies, practical training and human judgment to make sure the outputs are accurate, applicable and usable."
Why this matters for educators
For educators, the survey's findings underscore the importance of building accessibility into AI training and curriculum design. As AI tools enter classrooms, the ability to create accessible materials and evaluate AI outputs for accuracy becomes a concrete skill. 90% of respondents identified AI and accessibility skills as valuable for their current or desired career field, with 45% specifically seeing value in knowing when AI-generated content needs human review. The AI Learning Path for Teachers provides structured guidance on integrating AI responsibly in educational settings, including accessible design principles. Educators who can pair AI tools with clear accessibility practices will be better positioned to support all learners-and to model the judgment that AI alone cannot supply.
Your membership also unlocks:
