Building Accessible UI using GitHub Copilot
summary
Demo uses GitHub Copilot to retrofit an Adobe Express thumbnail multi-select: first change is a div→button, then Copilot suggests ARIA and roles, and after 4–5 iterations the screen reader finally announces labels and selection state. A remaining miss for a disabled thumbnail is solved by adding aria-disabled fix. Key claims: Copilot speeds suggestions but shows false confidence and needs clear, specific prompts and reliable sources (WCAG WAI-ARIA MDN). Lesson: iterative prompts plus human screen‑reader testing produce a working component.
AI-written summary · about
chapters (6)
- 0:00Introduction and speaker intro
- 1:48Understanding accessibility barriers
- 9:05Building accessible UI with GitHub Copilot
- 14:03Crafting effective prompts for AI
- 22:38Challenges and solutions in AI accessibility
- 24:25Human role in accessible design
related talks
31mCase Study: Building Accessible UI using Github Co-pilot
49mEmpowering People with Disabilities Using GitHub Copilot
32mAccessibility powered by AI
31mDeveloping for Everyone
40mBuilding Accessible Reusable Components
39m”Garbage Pail Components”
27m2024 09 20 06 ALENA NIKOLAEVA & DAVID JENSENIUS -The A1lyboros maximizing accessibility with Copilot
47mBuilding Accessible Web Components