Abstract
Mathematicians and engineers of the 1940s set out to design machines capable of making critical decisions in the face of uncertainty. In less than a decade, they developed a suite of powerful mathematical technologies, including information theory, linear programming, game theory, and neural networks. These tools didn’t merely lay the foundation for contemporary computer science but became the foundation of a new definition of rationality itself, reshaping how we judge human decision making. The Irrational Decision traces how a computational framework designed for machines came to define rationality for people in economics, public policy, and popular culture.
This talk will explore success stories in accelerating computers, regulating pharmaceuticals, and deploying electronic commerce. The successes reveal a common pattern. Automated decision systems thrive in well-defined spaces with clear rules and measurable goals. However, the same systems produce results ranging from brittle to absurd outside those controlled settings. Given these strengths and limitations, the talk will close by asking how we can put these tools to work without surrendering human agency and judgment.
Bio
Benjamin Recht is a Professor in the Department of Electrical Engineering and Computer Sciences at the University of California, Berkeley. Ben studies the foundations of machine learning, be they mathematical, statistical, and computational, or philosophical, sociological, and historical. He is the author of three books: Optimization for Modern Data Analysis (2022) with Stephen J. Wright, Patterns, Predictions, and Actions (2022) with Moritz Hardt, and The Irrational Decision (2026). He currently spends too much time blogging at argmin.net
Orchard View Room
Ben Recht, UC Berkeley