Kai Zhu

This paper studies the deployment of neural machine translation across more than 100 Wikipedia language communities and uncovers what we term the “AI Democratization Paradox,” where the technology simultaneously drives democratizing and concentrating forces. AI lowered barriers, leading to a substantial increase in content creation across diverse target languages without sacrificing quality or readership. However, the benefits were concentrated: well-resourced communities captured disproportionate gains—3–4 times larger than mid-tier editions. While editors actively leveraged AI to address representation gaps, translating female biographies at twice the expected rate, structural constraints limited the impact in high-need areas.

This paper investigates the consequences of monetizing a marketplace for product promotion within a digital platform, specifically the Giveaways program on Goodreads.com. Using a natural experiment and fine-grained platform data from 2016 to 2020, we examine how introducing a fixed entry cost for content creators affected both supply and demand. We find a substantial decrease in overall program participation, particularly among indie publishers and self-published authors, leading to increased market concentration. We also document reduced genre diversity, with popular genres becoming more dominant at the expense of niche categories, and intensified promotional effects characterized by higher review volume but lower average ratings.

This paper investigates how negative peer feedback shapes user behavior in an online review platform. Leveraging fine-grained digital trace data of users throughout their tenure on the platform, we find that negative feedback improves user retention on the platform relative to users who receive no feedback at all. For users who are retained and write their next reviews, negative feedback improves review frequency and quality. These findings demonstrate that, contrary to conventional wisdom, negative peer feedback can benefit the platform’s welfare as it increases user engagement and retention, and that no feedback can be a worse alternative.

Throughout the COVID-19 crisis, as we confronted questions about social distancing, mask wearing, and vaccines, public safety experts warned that the consequences of a misinformed population would be particularly dire. Through automated text analysis of complete transcripts of national cable, network, and local news, we explore their narratives surrounding the COVID-19 pandemic and characterize the differences in which topics were covered and how they were covered by various media sources. Our analysis reveals polarized narratives around blame, racial and economic disparities, and scientific conclusions about COVID-19.

Open collaboration platforms have fundamentally changed the way that knowledge is produced, disseminated, and consumed. We leverage a large-scale natural experiment to study how exogenous content contributions to Wikipedia articles affect the attention that they attract and how that attention spills over to other articles in the network. Results reveal that exogenously added content leads to significant, substantial, and long-term increases in both content consumption and subsequent contributions. Furthermore, we find significant attention spillover to downstream hyperlinked articles.

Generative AI depends on high-quality web content, yet no market compensates its producers. We document adverse selection in this AI data commons: facing a binary opt-out choice, the highest-quality producers exit first, degrading the remaining commons. Studying media and news sites at scale, we find a steep quality-blocking gradient: high-factual outlets block at nearly six times the rate of low-factual sources, with misinformation sources remaining most accessible. Publishers strategically target training crawlers while blocking search crawlers at lower rates. Event studies trace this divergence to the introduction of AI-specific opt-out mechanisms, and simulations show the quality degradation stems from the compositional pattern of exit. Current opt-out regimes produce adverse selection by design; our findings call for market mechanisms that properly compensate content producers.

This paper studies how platforms are increasingly offering direct monetization opportunities through creator-viewer subscriptions, contrasting with traditional reach-focused advertising models. We examine the impact of the Twitch Affiliate Program, which significantly lowered entry barriers to paid channel subscriptions and micro-donations. Using a time-shifted difference-in-differences design, we find that the program substantially increased content supply, with smaller creators more than doubling their streaming hours. Monetization access professionalizes creator behavior, leading to more deliberate content strategies and improved performance.

We develop a novel representation learning framework to quantify the fit between consumers and products using large-scale observational data. Our approach learns latent representations that capture nuanced preference patterns beyond traditional collaborative filtering. We demonstrate the method on book consumption data and show that our learned representations substantially improve prediction of consumer choices and reveal interpretable dimensions of consumer-product compatibility.