Ep 24: Solon Barocas on Removing Bias From Machines

Machine Learning can improve decision making in a big way -- but it can also reproduce human biases and discrimination. Solon Barocas joins Vasant Dhar in episode 24 of Brave New World to discuss the challenges of solving this problem.  Useful resources: 1.  Solon Barocas at his website, Cornell, Google Scholar and Twitter. 2. Fairness and Machine Learning -- Solon Barocas, Moritz Hardt and Arvind Narayanan. 3. Danger Ahead: Risk Assessment and the Future of Bail Reform: John Logan Koepke and David G. Robinson. 4. Fairness and Utilization in Allocating Resources with Uncertain Demand -- Kate Donahue and Jon Kleinberg. 5. Profiles, Probabilities, and Stereotypes -- Frederick Schauer. 6. Thinking Like a Lawyer: A New Introduction to Legal Reasoning -- Frederick Schauer. 7. Measuring the predictability of life outcomes with a scientific mass collaboration -- Mathew Salganik and others. 8. Inherent Trade-Offs in the Fair Determination of Risk Scores -- Jon Kleinberg, Sendhil Mullainathan and Manish Raghavan. 9. Limits to Prediction -- Arvind Narayanan and Matthew Salganik. 10. The Fragile Families Challenge. 11.  Daniel Kahneman on How Noise Hampers Judgement -- Episode 21 of Brave New World. 12.  Noise: A Flaw in Human Judgment -- Daniel Kahneman. 13. Dissecting "Noise" — Vasant Dhar. 14. Nudge: Improving Decisions About Health, Wealth, and Happiness -- Richard Thaler and Cass Sunstein.