Doing Data Science with Ethics
summary
Argues for concrete rules: register ethics goals at design time, track data provenance and consent, and treat fairness, interpretability and robustness as interconnected engineering requirements. Cites IBM facial‑recognition higher error rates for women and people of color, Strava revealing military sites, and YouTube optimizing clicks that amplifies polarization. Recommends tools like partial dependence plots, counterfactual probes, automated audits and privacy techniques (federated learning, privacy‑first search) and stresses documenting tradeoffs and costs of false positives.
AI-written summary · about
chapters (13)
- 0:00Start
- 0:07Making Data Science Work
- 0:50Doing Data Science with Ethics
- 7:08Regulatory Landscape
- 12:40Possible Ways That Devices in Your Home Can Be Hacked
- 29:07Representational Harm
- 33:41Case Studies
- 38:04The Early Phase of the Data Science
- 45:02Fairness of Prediction Systems
- 49:35Algorithm Engineering
- 50:39What Is the Cost of Going Wrong
- 51:16Interpretability
- 52:24The Difference between a Machine Learning System and a Human Brain
related talks
31mEthics of algorithms - from theoretical discussion to practical application
31mEthics and Data Science
47mEthics & AI: Identifying & preventing bias in predictive models
47mAI Ethics, Strategic Decisioning and Game Theory
31mPanel Discussion : Ethics in Data Science
47mThe Ethics of Software - some practical considerations
56mEthics in Machine Learning Panel
42mInfosec, AI and ethics – new models for a secure future