Sign up to save your podcasts
Or
Share Are ChatGPT, Claude, and Gemini Accessible? A Revealing Conversation with Joe Devon
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
Are ChatGPT, Claude, and Gemini Accessible? A Revealing Conversation with Joe Devon
In this in-person episode of Investing in Accessibility, Kelvin Crosby and Chris Maher are joined by returning guest Joe Devon—accessibility pioneer, GAAD co-founder, and longtime technologist—for a deep dive into one of the most urgent questions of our time: Is artificial intelligence being built accessibly?
Joe introduces AIMAC (AI Model Accessibility Checker), a new benchmark designed to measure how accessible AI-generated code really is—and to force AI companies to compete on inclusion, just as they do on speed, accuracy, and intelligence.
What AIMAC uncovered is both surprising and revealing.
When Joe tested 36 leading AI models, the results challenged common assumptions across the industry. Google’s Gemini, despite excelling in traditional AI benchmarks and having extensive accessibility tools at its disposal, performed poorly on accessibility. Anthropic—often positioned as a leader in ethical and responsible AI—delivered middle-of-the-road results, raising questions about how ethics translate into inclusive outcomes. In contrast, OpenAI’s ChatGPT models consistently led the pack, a result Joe attributes to intentional focus on accessibility and sustained engagement with the disability community. The takeaway is clear: when accessibility is prioritized, it shows up in the results.
Beyond rankings, the conversation explores why inaccessible AI coding tools—especially command-line AI interfaces—pose a serious risk of locking people with disabilities out of the future of work and innovation. At the same time, Joe, Kelvin, and Chris make the case that accessible AI represents a massive opportunity: for better technology, stronger businesses, and real empowerment for disabled entrepreneurs and creators.
This episode is a call to lean in, demand table-stakes access, and ensure the AI revolution does not repeat the exclusionary mistakes of the past.
If AI is shaping the future, accessibility must shape AI.
Links & Resources:
AIMAC (AI Model Accessibility Checker): Website
Joe Devon: LinkedIn
COMING SOON!
American Sign Language (ASL) and Captioning for each episode will be provided on our YouTube channel. Go to handle @SamaritanPartners.
View all episodes
By Kelvin Crosby & Chris Maher
5
1010 ratings
In this in-person episode of Investing in Accessibility, Kelvin Crosby and Chris Maher are joined by returning guest Joe Devon—accessibility pioneer, GAAD co-founder, and longtime technologist—for a deep dive into one of the most urgent questions of our time: Is artificial intelligence being built accessibly?
Joe introduces AIMAC (AI Model Accessibility Checker), a new benchmark designed to measure how accessible AI-generated code really is—and to force AI companies to compete on inclusion, just as they do on speed, accuracy, and intelligence.
What AIMAC uncovered is both surprising and revealing.
When Joe tested 36 leading AI models, the results challenged common assumptions across the industry. Google’s Gemini, despite excelling in traditional AI benchmarks and having extensive accessibility tools at its disposal, performed poorly on accessibility. Anthropic—often positioned as a leader in ethical and responsible AI—delivered middle-of-the-road results, raising questions about how ethics translate into inclusive outcomes. In contrast, OpenAI’s ChatGPT models consistently led the pack, a result Joe attributes to intentional focus on accessibility and sustained engagement with the disability community. The takeaway is clear: when accessibility is prioritized, it shows up in the results.
Beyond rankings, the conversation explores why inaccessible AI coding tools—especially command-line AI interfaces—pose a serious risk of locking people with disabilities out of the future of work and innovation. At the same time, Joe, Kelvin, and Chris make the case that accessible AI represents a massive opportunity: for better technology, stronger businesses, and real empowerment for disabled entrepreneurs and creators.
This episode is a call to lean in, demand table-stakes access, and ensure the AI revolution does not repeat the exclusionary mistakes of the past.
If AI is shaping the future, accessibility must shape AI.
Links & Resources:
AIMAC (AI Model Accessibility Checker): Website
Joe Devon: LinkedIn
COMING SOON!
American Sign Language (ASL) and Captioning for each episode will be provided on our YouTube channel. Go to handle @SamaritanPartners.