TY - JOUR

T1 - Complementarities between algorithmic and human decision-making

T2 - The case of antibiotic prescribing

AU - Ribers, Michael Allan

AU - Ullrich, Hannes

N1 - Funding Information: We benefited from valuable feedback by the editor and anonymous referees and helpful suggestions by Jason Abaluck, Rolf Magnus Arpi, Lars Bjerrum, Chiara Canta, Gloria Cristina Cordoba Currea, Greg Crawford, Tomaso Duso, G\u00FCnter Hitsch, Shan Huang, Ulrich Kaiser, Reinhold Kesler, Jenny Dahl Knudsen, Sidsel Kyst, Chlo\u00E9 Michel, Jeanine Mikl\u00F3s-Thal, Maria Polyakova, Carlo Reggiani, Sherri Rose, Stephen Ryan, Karl Schmedders, Aaron Schwartz, Andr\u00E9 Veiga, participants at the Annual Health Econometrics Workshop 2018, the 2019 CESifo Area Conference on the Economics of Digitization, the Digital Economy Workshop 2019, the 2019 NBER Conference on Machine Learning in Health Care, the International Conference on Computational Social Science 2020, as well as in seminars at DIW Berlin, ESMT Berlin, Toulouse Business School, University of Copenhagen, University of Zurich, and Vienna University of Economics and Business. Financial support from the European Research Council (ERC) under the European Union\u2019s Horizon 2020 research and innovation programme (grant agreement no. 802450) is gratefully acknowledged. Publisher Copyright: © The Author(s) 2024.

PY - 2024

Y1 - 2024

N2 - Artificial Intelligence has the potential to improve human decisions in complex environments, but its effectiveness can remain limited if humans hold context-specific private information. Using the empirical example of antibiotic prescribing for urinary tract infections, we show that full automation of prescribing fails to improve on physician decisions. Instead, optimally delegating a share of decisions to physicians, where they possess private diagnostic information, effectively utilizes the complementarity between algorithmic and human decisions. Combining physician and algorithmic decisions can achieve a reduction in inefficient overprescribing of antibiotics by 20.3 percent.

AB - Artificial Intelligence has the potential to improve human decisions in complex environments, but its effectiveness can remain limited if humans hold context-specific private information. Using the empirical example of antibiotic prescribing for urinary tract infections, we show that full automation of prescribing fails to improve on physician decisions. Instead, optimally delegating a share of decisions to physicians, where they possess private diagnostic information, effectively utilizes the complementarity between algorithmic and human decisions. Combining physician and algorithmic decisions can achieve a reduction in inefficient overprescribing of antibiotics by 20.3 percent.

KW - Antibiotic prescribing

KW - Antibiotic resistance

KW - C53

KW - D83

KW - Human-machine complementarity

KW - I18

KW - I19

KW - L2

KW - M15

KW - Machine learning

UR - http://www.scopus.com/inward/record.url?scp=85197472872&partnerID=8YFLogxK

U2 - 10.1007/s11129-024-09284-1

DO - 10.1007/s11129-024-09284-1

M3 - Journal article

AN - SCOPUS:85197472872

JO - Quantitative Marketing and Economics

JF - Quantitative Marketing and Economics

SN - 1570-7156

ER -