AI and Accessibility: the Good, the Bad, and the Bollocks
summary
Covers what actually works and what fails: demos with Seeing AI and iOS18 give real-time description and five‑minute video audio descriptions, Meta/Ray‑Ban wearables promise hands‑free scene narration, and a three‑minute voice clone shows useful speech backup. Warns that image generation produces stereotyped, wrong disability images and that overlays (AccessiBe, UserWay) make false ADA claims with pending lawsuits — overlays don't work. Emphasises that generative code and image captions hallucinate — hallucinations are risky.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:55Defining AI and accessibility
- 3:40Discriminative AI in accessibility
- 9:11Generative AI benefits
- 19:54Image generation bias
- 22:42Dangers of accessibility overlays
- 28:04AI code generation needs
- 31:16Hallucinations in AI outputs
- 33:21Summary
related talks
30mAI and Accessibility: the Good, the Bad, and the Bollocks
48mAI and Disability: Progress or Exclusion?
55mBridging the Digital Divide: Empowering Accessibility Through AI Innovations
57mCan Gen AI make the world Disability inclusive?
50mAI as an A11y Ally
43mAI and Disability: Progress or Exclusion?
34mAI for Accessibility: Empowering Inclusive Digital Experiences
49mThe impact of AI on people with disabilities