How we infuse our biases in the products...
summary
Claims that algorithms encode designers' values and that products reproduce social bias are illustrated with specifics: Siri omits abortion clinics for years then gets fixed, Facebook real-name policy blocks Native Americans and drag performers, a Ubisoft survey stops when users select female, and Microsoft’s Tay turns abusive when released. Machine learning amplifies historical bias and filter bubbles. Design tactics that actually get tested here are game‑designer practices: build for players, anticipate cheating, and design feedback loops to find and correct bias.
AI-written summary · about
chapters (8)
- 0:00Start
- 3:44OUR EXPERIENCE OF REALITY IS EMBEDDED IN OUR ALGORITHMS
- 4:33ALGORITHMS ARE IMPACTED BY THE PEOPLE WHO MAKE AND USE THEM
- 6:33GHOST IN THE MACHINE LEARNING
- 7:55WE ARE BLIND TO OUR OWN BIASES AND THE IMPACTS ARE GRAVE
- 9:52YOU PRODUCT EXISTS FOR YOUR PLAYERS
- 10:21PREPARE FOR PEOPLE TO CHEAT, HACK, EXPLOIT LOOPHOLES
- 10:45PREPARE TO BE WRONG, SO DESIGN FOR FAILURE, FEEDBACK LOOPS AND LEARNING CURVES
related talks
41mDesign for Cognitive Bias
30mInside AI: What Every Professional Needs to Know to Stay Relevant, Protected, and Ahead
11mMachines of Love & Grace – Building Truly Inclusive AI
05mToward Ethical AI
38mDesign for Cognitive Bias: Using Mental Shortcuts for Good
46mHow does designing your culture help your code?
24mDesigning for Transparency in Machine Learning
24mHey you, do you even design ?