Research

Usage mandates can accelerate AI adoption, but they also risk turning adoption into a metric to be gamed. We study a medical-device company that required roughly 5,000 employees to submit 200 AI queries per month, then reduced the target to 100 through a staggered branch rollout. Complete query logs show that the mandate increased first-time adoption, but 32% of queries are either repeated or non-work queries under the 200-query target. Lowering the target reduced query volume by 30%, improved query quality, and increased sales by 7% among sales employees. Our findings suggest that the design of usage targets shapes both how employees use AI and the returns from that use.

We examine barriers to new technology adoption among software engineers. Digital traces from 28,698 engineers at a tech company show only 41% adopted a generative AI programming tool twelve months after its launch, despite company-wide incentives, with lower rates among female (31%) and mature-age (39%) engineers. A pre-registered experiment (N = 1,026) reveals a competence penalty on tech adopters: engineers using AI received 9% lower competence ratings for identical work, with females penalized more (13%) than males (6%). A survey (N = 919) finds that anticipated competence penalty is associated with slower adoption, with stronger effects among female and mature-age engineers. These findings reveal a paradox: technologies intended to enhance productivity may inadvertently undermine their users, impeding adoption and reinforcing inequality.

By 2024, TikTok and Douyin reached 2.6 billion users globally. We provide large-scale quasi-experimental evidence on how short-video platforms affect mental health, using China as a case study. We combine city-level Douyin usage data with the China Family Panel Studies (2010–2022) and instrument for platform exposure using pre-2016 variation in 4G infrastructure. A one-standard-deviation increase in city-level Douyin screen time reduces mental health by 0.14 standard deviations. Because identification leverages city-level variation, estimates capture the combined effect of direct exposure and peer spillovers, the policy-relevant parameter. We further exploit individual-level 2016 mobile internet usage to sharpen identification, showing that negative effects concentrate among pre-existing mobile users and are absent among non-users. Effects are broadly similar across age, income, and initial mental health conditions, education, and employment status, suggesting group-specific confounders are unlikely to drive the results. We document three channels: worsened sleep and health routines, deteriorated physical health and cognitive function, and intensified unfavorable social comparison.

We examine how gender bias is transmitted between generations of managers within the organization and its negative consequences. Exploiting novel measures of managerial gender bias based on the incentive allocation policy and quasi-random rotation of managers and workers in a large construction firm in China, we find that a one-S.D. increase in the average managerial gender bias of the projects that the manager experienced as a subordinate is associated with a 0.19-S.D. increase in the managerial gender bias in his later projects. Such effects reflect persistence of gender-related social norms at work that partly originates from background social norms of managers during formative years. In addition, such managerial gender bias negatively affects both individual performance of female workers and overall team performance.

This paper studies how commission rates affect app entry across platforms. We examine an increase in Android game commission rate from 30% to 50% in China in 2014 and its impact on Android and iOS app stores, separately, in a difference-in-differences framework. We find a direct negative entry effect on Android by 47%. Meanwhile, the number of new games on the iOS App Store significantly decreased by 30% due to the higher Android commission rate, implying a negative cross-platform spillover effect. Moreover, the share of high-quality new games significantly decreased by 12%, indicating that higher commission rates discourage developers’ quality provision.