Inclusive & Ethical NLP Design Building AI for Everyone
summary
Workshop explains concrete methods: health‑first accessibility checks, WCAG/UAAG compliance, and iterative QA per task; an immersive avatar interface raises one autistic learner’s retention by 250% improvement; Europe’s Accessibility Act threatens $1M fines for noncompliance in native code; LLMs are framed as next‑token predictors, transformers since 2017 enable context via attention‑based transformers; practical stack recommendations include GitHub, Python, open‑source datasets, and community‑led data collection for low‑resource languages.
AI-written summary · about
chapters (16)
- 0:00Inclusive NLP design intro
- 1:03Ignis mission for inclusion
- 3:12Hannah Klaus introduction
- 6:28Grace Hopper and inclusive coding
- 11:51Diversity beyond gender
- 15:09Health internet accessibility
- 20:12EU regulatory changes
- 22:06AI agents accessibility
- 23:12AI ethics foundations
- 27:07NLP basics and challenges
- 36:45LLM bias problems
- 45:38Fair inclusive language models
- 52:45How to build large language models for everyone
- 1:00:03Low-resource languages
- 1:08:34Call to action for AI
- 1:09:16Q&A emotion AI and AGI
related talks
49mUsing AI to Streamline Inclusive Design Without Losing Humanity
35mWrite inclusive, people-first language about AI
01h 02mAI for Inclusive Society - Responsible AI
50mHuman-Centered AI and Accessibility
48mProgramming for Accessibility
40mMicrosoft AI for Inclusion and Innovation
51mBalancing Accessibility, Security and AI: Design Inclusive Security Tools
22mNothing for Us, Without Us; Breaking Unconscious Bias in Building Products