Universal Design or Universal Barrier? Rethinking AI Infrastructure on Campus

Many colleges and universities continue to operate with fragmented support systems, where accessibility responsibilities are confined to specific offices or compliance units. This siloed approach inhibits a cohesive strategy for digital inclusion. For instance, instructional design teams may not coordinate with disability services or IT departments to ensure that AI-powered tools align with accessibility standards. As a result, a new advising chatbot or AI-based learning platform might launch without screen reader compatibility or support for text-to-speech, effectively excluding students who rely on those tools. A 2020 survey by EDUCAUSE found that only 38 percent of institutions had a centralized digital accessibility policy, illustrating a widespread lack of institutional alignment around inclusive technology deployment¹.

Outdated digital infrastructure adds another layer of difficulty. Legacy systems, particularly student information systems and learning management platforms, often lack APIs or integration capabilities needed to support adaptive AI tools. When these systems are not interoperable, AI cannot access the data required to personalize learning experiences or respond to support needs in real time. This disproportionately affects nontraditional learners who rely on flexible, tech-mediated pathways. Without updated infrastructure, even well-intentioned AI initiatives may perpetuate existing inequities by default rather than design.

How AI Exacerbates Inaccessibility Without Governance

Artificial intelligence does not operate in a vacuum. It reflects and amplifies the systems it is embedded in. When implemented in environments with inconsistent accessibility standards, AI tools can reinforce exclusion. For example, automated essay scoring tools trained on standard English syntax may penalize multilingual learners or neurodiverse students whose writing deviates from normative patterns. In the absence of human oversight and inclusive training data, these systems risk codifying bias rather than supporting learning².

Poor governance compounds the problem. Many institutions lack formal AI governance structures that include accessibility experts, student advocates, and faculty from diverse backgrounds. Without interdisciplinary oversight, decisions about AI deployment may prioritize efficiency or novelty over inclusivity. A study by the Center for Democracy & Technology found that only 26 percent of higher education institutions had evaluated the equity implications of AI tools before implementation³. This gap in governance leaves students with disabilities and other marginalized groups exposed to tools that were never designed with them in mind.

Designing Forward with Inclusive Infrastructure

Inclusive AI design begins with universal design principles, which aim to create environments usable by all people, to the greatest extent possible, without the need for adaptation. This means ensuring that all AI interfaces support screen readers, keyboard navigation, and alternative input modes such as voice commands or switch devices. These features must be built in from the outset, not added after accessibility complaints arise. Web Content Accessibility Guidelines (WCAG) 2.1 provide a technical foundation, but institutions must go further by integrating user testing with students from diverse ability and language backgrounds⁴.

Multilingual design is equally essential. AI-powered chatbots, tutoring platforms, and advising systems should support multiple languages and dialects, especially in institutions serving immigrant, refugee, and first-generation college populations. Real-time language translation, culturally responsive prompts, and the ability to toggle between languages can significantly expand access. AI tools must also be monitored by human staff trained in inclusive pedagogy to ensure that automated responses align with institutional values