Mitigating Human Bias in Decision Making through Artificial Intelligence: Empirical Evidence from a Professional Service Platform

Posted: 14 May 2023 Last revised: 5 Dec 2023

See all articles by Kejia Hu

Kejia Hu

Vanderbilt University - Operations Management

Bowen Lou

University of Connecticut - Operations & Information Management Department

Bilal Baloch

Independent

Date Written: May 2, 2023

Abstract

As artificial intelligence (AI) emerges as a new operational foundation of business, there is growing interest in leveraging AI to address operational challenges. AI holds the potential to mitigate subjective biases that arise from the human decision making involved in operations. In collaboration with a leading platform in the U.S. that provides extensive professional services to answer customers’ open-ended questions, we investigate whether AI-enabled text processing can reduce certain types of biases in human decision making and enhance the inclusion of diverse perspectives in answers. We empirically examine the effect of an exogenous AI introduction into the answer approval process of human agents. Our findings show that human agents tend to approve answers with a neutral tone and from answer contributors in higher positions of authority. AI, when properly developed to exploit past answer data, can mitigate the biases related to neutrality and authority. Furthermore, we find that the effect of AI to mitigate the biases is more pronounced when the past answers consist of more diverse perspectives in their respective question categories. Human agents tend to pay closer attention to the answers with higher approval scores offered by AI. AI is particularly beneficial for less experienced human agents when they make their answer approval decisions. Our analyses also demonstrate that answers with more inclusive perspectives enabled by the human agent-AI interactions receive higher customer evaluations. Implications of the findings for the design of AI systems and workflow of service operations are discussed. Overall, the empirical evidence provided in our study sheds light on the role of AI in overcoming human biases, enabling effective and inclusive service operations.

Keywords: bias, human-AI interaction, service operations, decision making

Suggested Citation

Hu, Kejia and Lou, Bowen and Baloch, Bilal, Mitigating Human Bias in Decision Making through Artificial Intelligence: Empirical Evidence from a Professional Service Platform (May 2, 2023). Available at SSRN: https://ssrn.com/abstract=4436554

Kejia Hu (Contact Author)

Vanderbilt University - Operations Management ( email )

Nashville, TN 37203
United States

Bowen Lou

University of Connecticut - Operations & Information Management Department ( email )

Bilal Baloch

Independent

Do you have negative results from your research you’d like to share?

Paper statistics

Abstract Views
543
PlumX Metrics