01
Literature Review
Building an understanding of digital accessibility, AI-based tools and existing research within the field.
AI & Digital
Accessibility
How are UX practitioners using AI to support digital accessibility — and where does it fall short?
Degree Project · 2026
01 · Overview
Can AI make digital experiences more accessible?
This degree project explored how UX practitioners perceive AI-based tools as support for digital accessibility within a user-centered UX design process.
The study examined both the opportunities AI may offer as a supporting tool and the limitations that emerge when automation meets the contextual and human-centered nature of accessibility work.
02 · Research Questions
How do UX practitioners approach the use of AI-based tools in accessibility work within the user-centered UXD process?
Subquestions
01
How do UX practitioners currently use AI tools in their work, and which organizational factors influence their use?
02
How do UX practitioners currently work with accessibility in the UXD process in terms of methods, routines and established practices?
03
How do UX practitioners perceive or envision that AI-based accessibility tools could affect the user-centered UXD process?
04
What opportunities, limitations and needs do UX practitioners identify in relation to AI-supported accessibility work?
03 · Research
02
Survey
Exploring how UX practitioners perceive and use AI when working with digital accessibility.
03
Interviews
Developing a deeper understanding of practitioners' experiences, reasoning and perspectives.
04
Analysis
Identifying patterns across the collected material through thematic analysis and descriptive statistics.
04 · Key Findings
AI can support accessibility work — but human judgment remains essential.
Across the survey and interviews, practitioners saw clear opportunities for AI to support accessibility work. At the same time, the findings revealed important limitations around reliability, context and human understanding.
01
AI works best as complementary support.
Practitioners were generally positive toward using AI in accessibility work, particularly for identifying issues, suggesting improvements and supporting repetitive tasks. Rather than replacing UX expertise, AI was primarily viewed as a complementary tool that could support and streamline existing workflows.
02
Efficiency does not equal human understanding.
AI can help identify accessibility issues more efficiently, but practitioners questioned its ability to understand context, individual needs and why an experience may feel difficult or exclusionary. This creates a tension between the efficiency offered by automation and the human-centered nature of UX and accessibility work.
03
Human judgment remains essential.
Concerns around hallucinations, bias, incorrect interpretations and limited contextual understanding created a need for caution when using AI-generated output. Practitioners emphasized the continued importance of user involvement, critical evaluation and professional responsibility when making accessibility decisions.
05 · Recommendations
Designing with AI, without designing out the human.
The findings suggest that AI can play a valuable role in accessibility work when it is used as a supporting tool rather than a replacement for human understanding and professional judgment.
01
Use AI as support, not authority.
AI can help identify issues, generate suggestions and reduce repetitive work, but its output should be critically evaluated rather than treated as a final answer.
02
Keep users in the process.
Automated tools cannot fully capture individual experiences, contexts and needs. User involvement remains essential for understanding whether an experience is genuinely accessible.
03
Make human review non-negotiable.
Accessibility decisions still require contextual understanding, professional responsibility and human judgment, particularly when AI-generated output may be incomplete or incorrect.
06 · Reflection
Accessibility cannot be automated away.
This project changed how I think about the role of AI in accessibility. I came to see AI not as a solution to accessibility problems, but as one part of a broader human-centered process — where context, critical evaluation and user involvement remain essential.