It's not a bug, it's a bias
summary
Argues that software encodes makers’ worldview so products reproduce systematic exclusions, citing Siri hides a body but misses abortion clinics as a concrete mismatch. Shows platform effects like negativity bias in reviews hurting small businesses and mitigation via Google/Yelp local‑guide systems. Explains ML danger: training on historic hires amplifies bias and scales it. Recommends concrete steps: monitor support and abandonment metrics, run checklists, diversify teams (a “random forest” of perspectives) and simplify docs for broader access.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:14Is Siri biased
- 2:45Can algorithms be biased
- 3:37What is an algorithm
- 4:57Procedural Rhetoric
- 6:56Normality
- 8:07Bad Mojo
- 9:06Yelp Google Maps
- 10:41Machine Learning
- 12:09What can we do
- 12:31Catch them
- 13:31Random forest
- 14:59Checklist
- 17:09Algorithms
related talks
38mBiases are bugs: algorithm fairness and machine learning ethics
20mThere is no such thing as bias-free tech. How models shape our world.
27mCrouching Harm, Hidden Bias—Addressing issues of fairness....
38mSocial Implications of Bias in Machine Learning
38mSocial Implications of Bias in Machine Learning
24mAlgorithms aren't neutral Making UX inclusive
35mBad computers, bad people
56mAn AI with an Agenda: How Our Cognitive Biases Leak Into Machine Learning