← Search

Does Machine Learning Automate Moral Hazard and Error?

Sendhil Mullainathan¹; Ziad Obermeyer²

¹ Harvard University, Littauer Center M-18, Cambridge, MA 02138 (e-mail: ) · ² Harvard Medical School, 75 Francis Street, Neville House, Boston, MA 02138 and Brigham and Women's Hospital (e-mail: )

Machine learning tools are beginning to be deployed en masse in health care. While the statistical underpinnings of these techniques have been questioned with regard to causality and stability, we highlight a different concern here, relating to measurement issues. A characteristic feature of health data, unlike other applications of machine learning, is that neither y nor x is measured perfectly. Far from a minor nuance, this can undermine the power of machine learning algorithms to drive change in the health care system--and indeed, can cause them to reproduce and even magnify existing errors in human judgment.

DOI

10.1257/aer.p20171084

Volume

107 (5)

Pages

476-480

Language

en

Export

BibTeX

Sources

crossref openalex