Aaron Roth, Professor of Computer Science at UPenn, discusses his recent book, The Ethical Algorithm: The Science of Socially Aware Algorithm Design.

On this episode of Technology & Prose, Aaron Roth joins host Nikita Aggarwal to talk about the importance of ethical algorithm design (1:34), differential privacy and designing privacy-preserving algorithms (6:09), the costs and trade-offs of designing for privacy (10:38), application of differential privacy by Apple, Google and the 2020 US Census (14:13), GANs and synthetic data (18:22), designing fair algorithms, the ProPublica COMPAS investigation, the value and challenge of formalising ethical norms (20:30), modelling fairness/accuracy trade-offs (28:26), multi-objective optimisation (32:36), algorithmic game theory and designing for cooperation (34:13), incentivising organisations to adopt ethical algorithm design, privacy and fairness regulation (40:28).

References

Dinur and Nissim, Revealing Information While Preserving Privacy

Goodfellow et al, Generative Adversarial Nets

Angwin et al, ProPublica report on COMPAS

Northpointe’s response to ProPublica

Kleinberg et al, Inherent Trade-offs in the Fair Determination of Risk Scores

Recorded on March 5th 2021.