January–February 2022
Article
Organization Science

Algorithm-Augmented Work and Domain Experience: The Countervailing Forces of Ability and Aversion

Format:Print

Abstract

How does a knowledge worker’s level of domain experience affect their algorithm-augmented work performance? We propose and test theoretical predictions that domain experience has countervailing effects on algorithm-augmented performance: on one hand, domain experience enhances a worker’s ability to accurately assess the quality of an algorithmic tool’s advice; on the other hand, highly experienced workers exhibit more aversion to algorithmic advice, relative to their own judgment. We exploit a within-subjects experiment in which corporate IT support workers were assigned to resolve similar problems both manually (using their own judgment) and using advice generated by an algorithmic tool. Relative to solving problems using their own judgment, we confirm an inverted U-shape between IT domain experience and performance for problems where the algorithmic tool generated advice. While low experience workers’ propensity to reject accurate algorithmic advice appears to be driven by lack of ability to accurately assess the algorithm’s advice, highly experienced workers appear to reject algorithmic advice due to algorithmic aversion.

Keywords

Automation; Domain Experience; Algorithmic Aversion; Experts; Algorithms; Machine Learning; Future Of Work; Employees; Experience and Expertise; Decision Making; Performance

Citation

Allen, Ryan, and Prithwiraj Choudhury. "Algorithm-Augmented Work and Domain Experience: The Countervailing Forces of Ability and Aversion." Organization Science 33, no. 1 (January–February 2022): 149–169. ("Best PhD Student Paper" at SMS conference 2020.)