Fight AI Bias With… Bias
summary
Argues for using controlled bias to correct biased systems and shows concrete techniques: pre/in/post‑processing on datasets and models, counterfactual tests, and constrained optimization that trades accuracy for parity. Uses real failures like Amazon’s same‑day rollout and predictive policing to motivate monitoring for traumatized data, and recommends tools and explainers — SHAP LIME integrated gradients, IBM aif360 and Fairlearn — with a warning that many libraries have rough edges.
AI-written summary · about
chapters (4)
- 0:00Start
- 5:40Components of Responsible use of Al
- 11:52Quantifying Fairness in Al
- 32:02Model behavior understanding is critical to fighting bias
related talks
56mAn AI with an Agenda: How Our Cognitive Biases Leak Into Machine Learning
31mFlagging *and* Fixing Bias in ML
27mCrouching Harm, Hidden Bias—Addressing issues of fairness....
35mZero Bias AI
23mRemoving Unfair Bias in the Machine Learning Pipeline
40mEthical AI: Addressing Bias & Algorithmic Fairness
38mSocial Implications of Bias in Machine Learning
47mEthics & AI: Identifying & preventing bias in predictive models