Your Algorithm isn't neutral...
summary
Concrete failures and fixes decide whether a product is inclusive: Apple HealthKit lacking reproductive tracking for a year, security logic that maps the title “Dr.” to men and locks out Dr. Louise Selby, voice models trained on narrow accents, photo filters that whiten faces, and Airbnb interfaces that enable racial filtering labeled #AirbnbWhileBlack. Practical responses include add friction to forms (Nextdoor lowers racist posts but raises completion time), build mapping apps for mobility and blind Tube navigation, and involve affected people from the start and cede final decisions when necessary.
AI-written summary · about
chapters (4)
related talks
24mAlgorithms aren't neutral Making UX inclusive
24mEliminating Unconscious Biases in Technology Through...
38mSocial Implications of Bias in Machine Learning
25mIt's not a bug, it's a bias
47mDiversity and Inclusion for Algorithmic Bias Considerations
47mCreating solutions for everyone
42mLet's Talk about AI, ML and Bias
12mHow we infuse our biases in the products...