AI Deep Dive: AI-Assisted Accessibility and Universal Design

Primary Audience: Disability services, instructional designers, IT/web teams, compliance officers, and faculty

In April 2026 the Department of Justice extended the ADA Title II digital accessibility deadline by a year — most institutions now face April 26, 2027 — but the extension changed only the dates. The standard (WCAG 2.1 AA), the scope, and the obligation are untouched. More telling for this session is why the DOJ moved the date: it concluded that current technology, generative AI included, cannot yet reliably automate accessibility remediation at scale, and that human oversight remains required — a point it made most forcefully about STEM content. That is the federal government validating, in the one domain where the stakes are legally enforceable, the principle running through this whole series: AI is a multiplier, not a fixer. This session meets institutions roughly six months out from the deadline they actually face, with a clear-eyed read on where AI genuinely accelerates the work and where it cannot carry it.

The tier logic maps cleanly here. Drafting alt text, generating first-pass captions, and producing multilingual communications are real Tier 1 augmentation — AI does this work well and saves meaningful staff time. But document remediation at scale, and especially complex STEM material, is exactly where the DOJ drew its human-oversight line: an automated remediation pass that’s confidently wrong doesn’t reduce legal exposure, it manufactures it, while producing a paper trail that looks like compliance.

The deeper point is governance. A remediation tool without a documented, governed program behind it is a brick without mortar — and the field has already made this pivot, moving from “buy an accessibility tool” to “build a risk-prioritized, auditable remediation program” with clear ownership across IT, disability services, communications, and procurement. Pairing automated scanning with human and user testing isn’t a nicety; counsel guidance treats it as the thing that catches what the tools miss. The institutions that will be defensible in 2027 are the ones building that program now, with AI as an accelerant inside it — not a substitute for it. The session will show current tools in action — AI-generated alt text in Microsoft 365 and Adobe Acrobat, real-time captioning in Zoom and Google Meet, document remediation and checking via Grackle and Adobe Acrobat’s Accessibility Checker, and multilingual communication tools — being precise throughout about where each genuinely helps and where human review remains required. This session is demonstration-driven — nothing to install. Bring a sample document, web page, or campus communication you’d like to assess for accessibility gaps.