"What was really surprising to us was how it [AI] was impacting the accessibility community," Colorado CIO David Edinger said in an interview at last month's National Association of State Chief Information Officers (NASCIO) Midyear Conference. "Folks with some kind of disability, particularly sight-type disabilities, were saying it was transformational in terms of how it was impacting their lives and their ability to work productively."
Experts like Minnesota's Chief Information Accessibility Officer Jay Wyant agree, finding that technologies like live transcription and translation have gotten "exponentially better" than they were when the COVID-19 pandemic broke out in 2020.
Wyant notes that AI also has potential to help software developers identify things they might have overlooked when striving to design accessible products.
Video transcript provided below.
Wyant and his colleague Marie Cohan, who has a similar title in the Texas Department of Information Resources, recommend a "trust but verify" posture when it comes to AI. The potential is vast, but manual testing is still needed.
"We can't rely on AI 100 percent," Cohan said. "It's really critical that we include people with disabilities first and foremost into design, development, testing, remediation."
Video transcript, Wyant:
Probably the biggest way it factors into our work today is through live transcription. So people like me, for example, going out to teams or doing calls or whatever, we're using the automated transcriptions when we have to. We have a lot of human translation as well but that's not always available and that has become exponentially better than when COVID started. So that's one thing where the AI has really made an impact in terms of accessibility. Going forward I see AI helping accessibility in terms of being an additional tool in the toolbox for developers and vendors who are building products out in that it helps them manage what they're doing. So, in other words, not so much that the AI are writing applications, but that the developer can design the application, develop components of the application, and the AI can help make sure they haven't missed anything, make sure they've included all the effective components for full accessibility. I think that's where the next step is.
The key is what tools are you using? How much can you trust those tools? With AI you must always trust but verify. Especially if you use it at the moment because you have no idea what the bias is from the people who develop that tool.
Video transcript, Cohan:
Across the industry though, AI has done a lot to help people with disabilities. There's a lot of applications out there now that do a number of different things to help a number of different disabilities. You've got, AI is now cropping up in a lot of the daily applications we use in business in the state and testing products, you know, educational products for people with disabilities or to train people like me for digital accessibility. Our testing tools, it's really showing up as far as the automated scanning.
I do caution though, we still need to do manual testing. We still need to test with assistive technology. We can't rely on AI 100 percent. It's really critical that we include people with disabilities first and foremost into design, development, testing, remediation. You know, there's a lot of potential for it but we also need to take baby steps with it too and make sure that we're not jumping ahead and not taking some things into consideration.