Accessibility powered by AI
summary
Claims that AI will replace accessibility work give context but not solutions instead the talk shows concrete experiments using GPT-3, ChatGPT and Github Copilot to generate an HTML CSS and JS prototype and then prompt the models to add larger fonts high contrast keyboard navigation alt text and simplified forms The demo code is entirely AI produced, Copilot injects changes into home.html, network hiccups occur during live run, and the takeaway is that explicit accessibility prompts produce useful patches but require human review and testing (eg color blindness ~12% of males noted).
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:02Why accessibility matters
- 3:12Diverse user needs
- 8:55Situational disabilities
- 11:51Age-related challenges
- 15:46Accessibility benefits all
- 16:29AI basics and limits
- 18:16Vibe coding risks
- 21:06AI for accessibility
- 21:49Live demo: AI accessibility
- 31:09Summary
related talks
28mYou Don’t Need to Worry About Accessibility with AI (Or Do You?)
48mProgramming for accessibility
47mAccessibility in the era of generative AI
41mProgramming for Accessibility
29mProgramming for Accessibility
25mBuilding Accessible UI using GitHub Copilot
43mHow to Use AI to Build Accessible Products
57mCan Gen AI make the world Disability inclusive?