AI Is Scaling Fast—Accessibility Isn’t. Here’s How We Fix That.

Guest: Joe Devon
Title: Chair, GAAD Foundation | Co-founder, Global Accessibility Awareness Day

AI is reshaping how we design software—but accessibility still too often shows up as an afterthought. In this episode of FUTUREPROOF., Joe Devon joins us to unpack what it actually means to build technology that works for everyone, especially as generative AI becomes embedded across products, platforms, and workflows.

Joe explains why accessibility isn’t a niche concern—it affects more than 1.3 billion people globally—and why AI represents both the biggest risk and the biggest opportunity the accessibility movement has ever seen. We dig into the early findings from the AI Model Accessibility Checker (AIMAC), what most AI models still get wrong about accessible code, and why “AI will fix it later” is a dangerous assumption.

We also explore how front-end tools like AI-generated captions, voice interfaces, and image descriptions are changing daily life for users with disabilities—and where back-end AI systems can finally close the gap between automated testing and real-world usability. Throughout the conversation, Joe makes a compelling case that accessibility is not just a moral imperative, but a design discipline that will separate future-proof products from legacy ones.

Topics covered:

• Why most digital products still fail basic accessibility standards
• How AI can dramatically expand—or quietly restrict—access
• What AIMAC reveals about how accessible today’s AI models really are
• Front-end vs. back-end accessibility breakthroughs
• The ethical stakes of deploying inaccessible AI at scale
• Why inclusive design must be a core requirement, not a patch