Creating solutions for everyone
summary
Shows concrete failures and fixes: real studies like seatbelts increase serious injury for women 47–71%, algorithms that mislabel risks (ProPublica finds black defendants flagged nearly twice as often), commercial systems that broke (Microsoft’s Tay shut down in ~16 hours, Amazon retires biased hiring AI), and product changes that work (Spain’s organ-donor default raises consent to ~99%, Microsoft Accessibility Insights and UK GOV design patterns). Covers form fixes (accept full name, optional gender), abuse threat models (STRIDE) and examples where defaults or design choices harm users.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 2:05My own experiences
- 8:12Form design defaults
- 15:14What should we do
- 20:24Edge cases
- 25:49Security model
- 33:49Preventing abuse
- 41:49Resources
- 43:49Microsoft advert
related talks
41mDesign for Cognitive Bias
47mSystems of Harm
38mDesign for Cognitive Bias: Using Mental Shortcuts for Good
24mDesigning for Transparency in Machine Learning
24mHey you, do you even design ?
24mEliminating Unconscious Biases in Technology Through...
23mYour Algorithm isn't neutral...
30mDesign in an Age of inclusivity