Machine Intelligence, Bureaucracy, and Human Control

Abstract

AbstractMachine intelligence, used extensively throughout modern bureaucracies, is quickly evolving, giving rise to machine agents that accomplish tasks typically reserved for human beings. This shift affects task completion, human/machine coproduction, and the control of the bureaucracy itself. Using Max Weber’s ideal type bureaucracy as a guiding construct, we argue that machine agents may offer technical capacity for task completion beyond that of humans. Furthermore, the technical strengths of machine intelligence, including (1) speed, (2) dispassion, (3) predictability, and (4) rational rule-based functioning, mirror those found within Weber’s ideal type. Through this lens, the evolution of both bureaucratic structures and the decision-making agents within them presents at least three important challenges for modern scholars: (1) deciding the scope of tasks machine agents should complete, (2) adapting the bureaucracy to work with machine agents, and (3) retaining the role of humans and human control.