Whitman Voices

Introduction

Does Machine Learning Pose a Risk to the Benefits of Sentient Human Decision Making in Organizations?

Does Machine Learning Pose a Risk to the Benefits of Sentient Human Decision Making in Organizations?

It is well accepted that machine learning benefits organizations in terms of speed and the volume of information that can be processed. Businesses, hospitals, schools and countless other organizations have come to rely on machine learning algorithms to make decisions that impact the way they operate. While this is, no doubt, beneficial and efficient, can it really be a substitute for the value of intuitive human decision making within an organization? 

A study, “Substituting Human Decision Making and Machine Learning: Implications for Organizational Learning,” by Natarajan Balasubramanian, Ph.D., professor of management at Syracuse University’s Martin J. Whitman School of ManagementYang Ye, Ph.D., associate researcher, Southwestern University of Finance and Economics, Chengdu, China; and Mingtao Xu, Ph.D., assistant professor, E.J. Ourso College of Business, Louisiana State University, asserts that machine learning risks lowering the extent of diversity and background knowledge in organizational routines. This, the authors argue, can increase the risk of learning myopia in organizations. 

“By laying out the important contingencies that affect the trade-off between the benefits of machine learning and the benefits of routine diversity and knowledge richness, we hope to enable scholars and practitioners to achieve a deeper understanding of the organization learning-related risks due to machine learning,” says Balasubramanian.

The study argues that humans have the ability to develop diverse decision-making patterns by engaging with their environments and applying substantive rationality — something machines simply cannot do despite their speed and flexibility. It goes on to reiterate that because machine learning relies only on statistical analysis based on historical data that can arise from variations in routines to make decisions, it is not true sentient learning. 

To mitigate this risk, Balasubramanian and his co-authors emphasize the need for “strong governance mechanisms that focus on routine diversity and knowledge richness, and enable building alternative inventories of such diversity and knowledge that may be lost when machine learning replaces human decision making.”

The study is scheduled be published in the Academy of Management Review in 2021.

Learn more about research being completed by the faculty at the Whitman School.