This study investigates the effects of artificial intelligence (AI) adoption in organizations. We ask: (1) How should a principal optimally deploy limited AI resources to replace workers in a team? (2) In a sequential workflow, which workers face the highest risk of AI replacement? (3) How does substitution with AI affect both the replaced and non-replaced workers' wages? We develop a sequential team production model in which a principal can use peer monitoring—where each worker observes the effort of their predecessor—to discipline team members. The principal may replace some workers with AI agents, whose actions are not subject to moral hazard. Our analysis yields four key results. First, the optimal AI strategy involves the stochastic use of AI to replace workers. Second, the principal replaces workers at the beginning and at the end of the workflow, but does not replace the middle worker, since this worker is crucial for sustaining the flow of information obtained by peer monitoring. Third, the principal may choose not to fully exhaust the AI capacity at her discretion. Fourth, the optimal AI adoption increases average wages and reduces intra-team wage inequality.
We introduce a theory of dominance among choices that encompasses strict and weak dominance among strategies in games, Blackwell dominance among experiments, and first or second-order stochastic dominance among monetary lotteries as special cases. One choice dominates another if in a variety of situations the former choice yields higher expected utility than the latter. We show that under certain assumptions all undominated choices are optimal in some situation. We show this result for three different definitions of dominance. Our analysis of existing dominance notions in economics in one common framework allows us to compare the properties of these concepts, and to obtain insights into why certain versions of our result apply only to some, but not all of these concepts. We motivate the concept of undominated strategies by developing a formal result relating undominated strategies to rational inattention.
Two imperfectly informed experts are hired to advise a decision maker. The experts are assumed to report their private information truthfully. In this paper we compare the informativeness of different joint (conditional on the true state) distributions of the experts' private signals, keeping the conditional marginal distribution of each expert's private signal given and fixed. Our comparisons use Blackwell's (Blackwell, 1951) notion of informativeness. We interpret "diversity" as an absence of perfect correlation among experts' signals. Such diversity manifests itself in a positive probability that the experts disagree on which state of the world is more likely the true state. We find that joint distributions in which experts disagree more frequently often have an advantage over distributions in which disagreement is observed rarely. Disagreement may thus be a manifestation of beneficial diversity.
This paper investigates how a sender, by providing free information to a perfectly rational receiver, can manipulate the receiver's learning (change the receiver's learning outcome). Formally, I employ a Bayesian persuasion model with the additional assumption that the receiver has costly opportunities to acquire further information after receiving the information from the sender. It seems that the sender should make the receiver end up knowing more than she would otherwise (I call this "encouragement"). However, as shown in the paper, the sender may also make the receiver end up knowing less (I call this "deterrence") or knowing different aspects of the object than she would otherwise (I call this "diversion"). I identify the necessary conditions for the feasibility of these manipulations, where two properties of the receiver's information acquisition cost function, that the recent literature on information acquisition calls Sequential Learning Proofness (SLP) and the more restrictive Indifference to Sequential Learning (ISL), play a vital role.
写歌的人假正经，听歌的人最无情。