Data Science: a Critical Introduction

In 2012, Harvard Business Review named data science the sexiest job of the 21st century. Companies like Google and Netflix were celebrated as models of the “data-driven” organization, using data science and machine learning to improve decision-making, drive efficiencies, and serve content tailored specifically to each delighted user. And with the adoption of “Artificial Intelligence,” data products promise even greater efficiency and satisfaction. Yet, for data scientists themselves, the most pressing question is: Why do data projects so often fail? Datasets are routinely exposed as biased or noisy; and much-touted prediction systems often prove dangerously inaccurate. But what is data science? What counts as data? And how is data used—and misused—in everything from statistics to machine learning to product design and advertising?

In this course, we will examine the practice of data science in order to understand its limits and potential. Through the clarifying lenses of history, epistemology, philosophy of science, and metascience, we will sharpen our judgment about when we can expect data science to provide useful knowledge or predictions and when we can’t. We will ask: In what senses are data and data science ever objective? Is statistics obsolete? What is machine learning and what does it have to do with artificial intelligence? What’s in these “black-box” algorithms? What good are predictions without a theory? What can the history of data science tell us about the current moment in AI? Readings will draw from work by Sabina Leonelli, Heather Douglas, Hannah Arendt, Michel Foucault, Donna Haraway, Nick Srnicek, William Cleveland, John Tukey, Hilary Mason, Andrew Gelman, Chris Wiggins, and Timnit Gebru, among others.

Course Schedule

Wednesday, 6:30-9:30pm ET
November 13 — December 11, 2024
4 sessions over 5 weeks
Class will not meet Wednesday, November 27th

$335.00

This class is completed. Check out our upcoming courses.