Digital accessibility has moved from a compliance-oriented add-on to a core quality criterion for online and blended learning. Yet, recent reviews of Artificial Intelligence in Education (AIED) show that much work still concentrates on cognitive outcomes, STEM domains, and technical architectures, with comparatively limited attention to accessibility, equity, and the lived experiences of diverse learners (Mustafa et al., 2024; Wang et al., 2024). Reviews of digital accessibility and AI likewise caution that technical innovation often outpaces attention to disability, multilingualism, and socio-economic constraints (Chemnad & Othman, 2024; López-Gazpio, 2025).
This paper reports a design case from LDT 508: Design of Accessible Digital Learning, a fully online, six-week graduate course in Arizona State University’s Learning Design and Technologies (LDT) program. Taught in Summer 2025, the course integrates Web Content Accessibility Guidelines (WCAG 2.2), Universal Design for Learning (UDL 3.0), and generative AI (GenAI) tools to prepare future learning designers to create accessible, inclusive digital learning experiences. We describe the design model, including empathy challenges, AI-supported iterative design, and adaptation for low-bandwidth, mobile-first contexts in the Mastercard Foundation (MCF) Scholars Program. We then situate this model within current AIED and accessibility research and derive implications for instructional design practice and future research.
Research in AIED has documented learning gains from intelligent tutoring and adaptive systems (Wang et al., 2024) while also showing that research is heavily concentrated in STEM fields and framed around individual cognitive performance (Chen et al., 2020). Research also emphasizes that educators still lack clear design models for leveraging AI in ways that are pedagogically meaningful, equitable, and aligned with broader institutional strategies (Mustafa et al., 2024).
Recent work begins to shift this focus toward inclusion. GenAI can be integrated into accessible learning environments to scaffold comprehension, translation, and text simplification, while warning that these tools inherit biases and accessibility limitations from their training data and interfaces (López-Gazpio, 2025).
Accessibility and UDL scholarship provide robust frameworks for proactive, inclusive design (CAST, 2024). Digital accessibility and UDL can be combined to support multilingual learners with disabilities, emphasizing multiple means of engagement, representation, and expression grounded in WCAG-aligned practices (Shyyan et al., 2025). Studies of MOOCs and online programs show that when accessibility is treated as a retrofit rather than a design premise, learners with disabilities and those in low-resource contexts encounter persistent barriers in navigation, media, and assessment (Iniesto et al., 2022).
GenAI adds a new layer to this landscape. Conceptual and empirical work on GenAI in education documents opportunities for personalization, multimodal feedback, and automation of routine authoring tasks, alongside concerns about misinformation, bias, privacy, and “metacognitive offloading” when learners over-rely on AI (Jin et al., 2025). Stefaniak and Moore’s (2024) work on design deliberation is especially relevant: it positions GenAI as a catalyst for ethical, context-sensitive design deliberation rather than as an automatic solution.
Across these bodies of work, several gaps emerge: (a) a relative scarcity of design-based accounts that center accessibility and inclusion as primary aims; (b) limited integration of GenAI with operational accessibility frameworks like WCAG and UDL; and (c) insufficient attention to low-bandwidth, mobile-first contexts. LDT 508 was intentionally designed to address these gaps.
LDT 508 is a six-week, fully asynchronous graduate course in the MEd in the LDT program. The course enrolls aspiring learning designers and technologists, including cohorts in the MCF e-Learning Initiative focused on African institutions and low-resource environments. The primary objective is for students to design, implement, and evaluate learning experiences that:
align with WCAG 2.2 success criteria relevant to text, images, media, structure, and interaction.
embed UDL 3.0 principles by offering multiple means of engagement, representation, and action/expression.
use GenAI tools as design partners while maintaining human oversight, ethical awareness, and critical evaluation.
The course is organized into six units, each combining targeted readings, an iterative design task, a discussion-based empathy challenge, and structured reflection. Students build a single Google Sites–based educational website over the term, progressively enhancing its accessibility and inclusivity.
The course’s design model deliberately integrates WCAG 2.2 criteria, UDL 3.0 principles, and GenAI tools as mutually reinforcing elements rather than separate add-ons. The GreenPath exemplar site anchors this model: each unit introduces new accessibility and UDL features on GreenPath, which students then mirror and adapt in their own projects.
Unit-level progression: Each unit centers on a design problem tied to a specific accessibility focus:
Unit 1 – Foundations of Disability, Accessibility, and Inclusive Design. Students frame the purpose and audience of their site, define accessibility goals, and discuss disability, accessibility, and accommodation as intersecting constructs rather than interchangeable labels.
Unit 2 – Inclusive Text. Students compose course texts using plain language, structure headings, lists, and links for screen-reader compatibility, and use GenAI to propose alternative wordings. They validate readability and structure with automated checkers (e.g., W3C Easy Checks, WebAIM).
Unit 3 – Accessible Visuals. Students design images with sufficient contrast, meaningful alt text, and redundancy between text and visuals. GenAI is used to generate visual content to complement text, to draft descriptions for tables, graphs, and images.
Unit 4 – Audio and Video Accessibility. Students create captioned videos and audio content, using auto-captioning and AI transcription as starting points, then correcting timing and domain-specific terminology.
Unit 5 – Evaluation and Peer Review. Using a custom WCAG–UDL checklist, students conduct peer accessibility reviews, combining automated reports with manual inspection. GenAI is used to evaluate for inclusive language, assess images and other visual content for representation and potential stereotypes, and review the use of colors and symbols for potential cultural misinterpretations and reliance on color.
Unit 6 – Reflection and Future Practice. Students complete a “Working Out Loud” blog post narrating their accessibility journey, including points where GenAI helped, hindered, or required careful verification.
Empathy challenges as structured perspective-taking: Each unit includes a simulation-based empathy task (e.g., dyslexia, color vision deficiency, hearing loss, limited mobility). Rather than treating these as standalone activities, empathy challenges are directly linked to design decisions: Students must document specific changes they make to their sites in response to the simulated experience. This structure aligns with accessibility training research showing that perspective-taking is most powerful when it is embedded in authentic redesign tasks rather than framed as abstract awareness (Shyyan et al., 2025).
GenAI as co-designer, not author: In LDT 508, generative AI is framed as a co-designer that accelerates routine work but never replaces human judgment. Students use AI to draft plain-language text, translations, summaries, jargon explanations, alt text, captions, transcripts, image descriptions, and accessible color palettes. They then revise all outputs against WCAG and UDL guidelines, correct errors, and ensure cultural and contextual fit. AI is also used to review language for bias and analyze representation in visuals, while its limitations such as hallucinations, shallow accessibility checks, and cultural blind spots are explicitly discussed. The course concludes with ethical guidelines: use AI for drafting, pair it with equity frameworks, and maintain transparency and human oversight.
Students document AI use in a short “AI design log,” verify outputs using manual checks or external tools, and critique AI suggestions through a UDL and equity lens. This aligns with Stefaniak and Moore’s (2024) emphasis on design deliberation, and with broader evidence that GenAI can be accessibility-enabling only when paired with strong human oversight and AI literacy.
Through the Mastercard Foundation Scholars Program e-Learning Initiative, the course model was adapted for cohorts designing courses in African higher education contexts characterized by intermittent connectivity, shared devices, and mobile-dominant access. The GreenPath Learning Hub provided a shared infrastructure for asynchronous materials and project work. Three design principles guided this adaptation:
Mobile-first accessibility. Students were required to test all content on smartphones with limited data plans, prioritizing responsive layouts, low-bandwidth media alternatives (e.g., transcripts and downloadable text over streaming video), and offline-friendly formats.
Tool pragmatism. While GenAI tools can ease accessible authoring, many are blocked, costly, or unstable in low-resource contexts. The course therefore emphasized portable practices such as prompt patterns for generating alt text or simple language usable across available institutional or open-source AI tools.
Contextualized UDL. Building on calls to localize AI-enabled instruction ethically (Stefaniak & Moore, 2024), students analyzed how disability, connectivity, and language interact in their own institutions, then tailored UDL strategies accordingly.
These adaptations respond directly to warnings from accessibility and MOOC research that high-bandwidth, desktop-centric design can entrench inequities even when nominal accessibility standards are met (Iniesto et al., 2022).
LDT 508: Design of Accessible Digital Learning offers a replicable design model for integrating WCAG, UDL, and GenAI in an asynchronous graduate course. The course responds directly to gaps identified in recent AIED and digital accessibility reviews through combining empathy-driven activities, AI-supported iterative design, and adaptation for low-resource environments. Rather than treating accessibility as a compliance checklist or AI as a novelty, the design model situates both within a broader commitment to equity, human-centered design deliberation, and global relevance.
For instructional designers and programs, the key message is straightforward: accessibility and GenAI integration should be taught together, with explicit attention to ethics, infrastructure, and context. For researchers, the course provides a concrete, data-rich environment in which to investigate how AI-enabled accessibility training shapes designers’ practices, learner experiences, and institutional cultures over time.
EdTech Archives