Make world less shit. NOW.
summary
Shows that automation can hide prejudice: a 1982–1996 admissions system reproduces past staff biases until a racial equality complaint exposes it. Points to business tools like TaskRabbit and CrowdFlower that bypass employment protections, and to Wikipedia’s success yet skewed participation with 9% of contributors identifying as female. Argues for concrete choices—what data fields are stored, whether a database is “people” or “users”, whether a service is built to protect or to exploit—and reports whether those designs actually worked.
AI-written summary · about
chapters (13)
- 0:00Intro
- 0:25No slides
- 0:40Ethics of work
- 1:50Humanlike robot companions
- 6:03Privacy
- 9:12Bypassing employment laws
- 10:43Wikipedia
- 12:49Freedom Principle
- 17:57Design
- 20:35Young people
- 21:36Real problems
- 22:23Everyday decisions
- 24:18Tools of the future
related talks
11mMachines of Love & Grace – Building Truly Inclusive AI
29mUnintended consequences of technology: Enabling NextGen discrimination
28mBreaking the Chains of Oppressive Software
31mCan you Build Algorithmic Justice?
38mSocial Implications of Bias in Machine Learning
27mCrouching Harm, Hidden Bias—Addressing issues of fairness....
31mWorks On My Machine, or the Problem is between Keyboard and Chair
24mAlgorithms aren't neutral Making UX inclusive