Machine Learning and Human Capital Complementarities: Experimental Evidence on Bias Mitigation

Prithwiraj  Choudhury; Evan  Starr; Rajshree  Agarwal

Article | Strategic Management Journal | August 2020

Machine Learning and Human Capital Complementarities: Experimental Evidence on Bias Mitigation

by Prithwiraj Choudhury, Evan Starr and Rajshree Agarwal

Abstract

The use of machine learning (ML) for productivity in the knowledge economy requires considerations of important biases that may arise from ML predictions. We define a new source of bias related to incompleteness in real time inputs, which may result from strategic behavior by agents. We theorize that domain expertise of users can complement ML by mitigating this bias. Our observational and experimental analyses in the patent examination context support this conjecture. In the face of “input incompleteness,” we find ML is biased towards finding prior art textually similar to focal claims, and domain expertise is needed to find the most relevant prior art. We also document the importance of vintage-specific skills and discuss the implications for artificial intelligence and strategic management of human capital.

Keywords: Machine learning; bias; Human Capital; Management; Strategy;

Format: Print

Find at Harvard Read Now

Citation:

Choudhury, Prithwiraj, Evan Starr, and Rajshree Agarwal. "Machine Learning and Human Capital Complementarities: Experimental Evidence on Bias Mitigation." Strategic Management Journal 41, no. 8 (August 2020): 1381–1411.

Goldman Sachs runs an annual internship for over 3,000 participants, spread across dozens of the firm's global offices. In 2020, the team brought all its resources to bear to transform the internship program into a fully virtual format in just a few short weeks. The new all-virtual internship faced challenges, but also benefited from unexpected opportunities. As the program ended, the team reflected on what worked, what they would change, and what the future of the internship program at Goldman Sachs could look like in 2021 and beyond.

GitLab is a software company that works “all remote” at the scale of more than 1,000 employees located in more than 60 countries. GitLab has no physical office and its employees can work from anywhere they choose. Any step of the organizational life of a GitLab employee (e.g., hiring, onboarding and firing) is performed remotely, except for a yearly companywide gathering. GitLab strongly relies on asynchronous coordination, allowing employees to work anytime they want. After highlighting some of the main practices implemented by GitLab to effectively work all remotely and asynchronously, I asked renowned organizational scientists their thoughts on this interesting case and to question the generalizability of the all remote asynchronous model. Understanding whether and under what conditions this model can succeed can be of guidance for organizational designers that are now considering different remote models in response of the COVID-19 shock and its aftermath.

We investigate the relationship between the presence of migrant inventors and the dynamics of innovation in the migrants’ receiving countries. We find that countries are 25 to 60 percent more likely to gain advantage in patenting in certain technologies given a twofold increase in the number of foreign inventors from other nations that specialize in those same technologies. For the average country in our sample, this number corresponds to only 25 inventors and a standard deviation of 135. We deal with endogeneity concerns by using historical migration networks to instrument for stocks of migrant inventors. Our results generalize the evidence of previous studies that show how migrant inventors "import" knowledge from their home countries, which translates into higher patenting in the receiving countries. We interpret these results as tangible evidence of migrants facilitating the technology-specific diffusion of knowledge across nations.

Faculty & Research

Machine Learning and Human Capital Complementarities: Experimental Evidence on Bias Mitigation

Abstract

Citation:

About the Author

More from the Author