Manish Raghavan: Understanding Societal Impacts through Machine Learning and Mechanism Design

Date: October 19, 2021 Speaker: Manish Raghavan (Harvard University) Title: Understanding Societal Impacts through Machine Learning and Mechanism Design: Automated Hiring as a Case Study Abstract: The increased use of algorithmic techniques to make socially consequential decisions has prompted concerted efforts to understand the societal impacts of algorithmic decision-making. Much of this work draws primarily upon techniques and insights from machine learning; however, as algorithmic systems grow more complex, we need to incorporate ideas from game theory and mechanism design in order to fully understand their impacts. In this talk, I'll examine how machine learning and mechanism design can come together to deepen our understanding of how algorithms impact society, using automated hiring as an example. After providing a brief overview of algorithmic hiring, I'll consider the issue of "algorithmic monoculture" in this context. This concern invokes analogies to agriculture, where a monocultural system runs the risk of severe harm from unexpected shocks. Here we show that the dangers of algorithmic monoculture run much deeper, in that monocultural convergence on a single algorithm by a group of decision-making agents, even when the algorithm is more accurate for any one agent in isolation, can reduce the overall quality of the decisions being made by the full collection of agents. Unexpected shocks are therefore not needed to expose the risks of monoculture; it can hurt accuracy even under "normal" operations, and even for algorithms that are more accurate when used by only a single decision-maker. Our results rely on minimal assumptions, and involve the development of a probabilistic framework for analyzing systems that use multiple noisy estimates of a set of alternatives.