Research

Working Papers

  • Is Generative AI A Complement or Substitute to Human Creativity? (JMP)
    Presentations: University of Toronto 2025
    How does generative AI impact human creativity? Unlike many other innovations, generative AI is being used to automate specific aspects of the creative process. This paper investigates a setting where I observe all intermediate steps of visual art creation using an AI image generator. I find that, over time, artists become more efficient in using generative AI: they converge on their desired output more quickly, use fewer stopwords, and incorporate more style-related terms in their prompts. Analyzing rich prompt-level data, I show that artists construct prompts strategically, gradually adding words to prompts and changing important words earlier in the process. Using large language models, I convert output images into text and embeddings, which allows me to study whether artists converge to similar patterns as their experience accumulates. This study aims to study how generative AI shapes human creativity over time.

  • Hiding From Generative AI
    Revise and Resubmit, Quantitative Economics
    Presentations: TSE Digital Economics Conference 2025; University of Toronto 2024; NASM-ES 2024; ESIF Economics and AI+ML 2024; EEA-ESEM 2024; ALEA 2024; CEA 2024
    Paper on SSRN | Slides
    How does generative Artificial Intelligence (AI) change the behaviors of content creators? I investigate the effect of an AI image generator on artists’ incentives to publish artworks using data from an online art platform, DeviantArt. On November 11 2022, DeviantArt introduced a generative AI image generator into the platform and artworks on this platform entered training data by default. Using a difference-in-differences estimation with artists who do not use AI, I show that digital artists publish 21% fewer artworks following AI’s introduction on this platform, in contrast to artisan crafts artists. This reduction could potentially hinder knowledge spillovers to other artists and AI training data availability. By matching the artworks of artists who publish both on DeviantArt and Instagram, I find that despite artists publishing fewer artworks on DeviantArt, the quality of published artworks for a given artist remains the same after the introduction of AI.

Work In Progress

  • Platforms in Platform
    Platforms could be part of a larger platform like an app store. Even though there exists a positive cross-group network externality between buyers and sellers, there is still an incentive for buyers and sellers to segment into different platforms. Buyers may choose to use different platforms to access more targeted products, while sellers may use different platforms to avoid competition for buyers’ attention. This paper argues that segmenting the market using smaller platforms can be profitable for the large platform, as it can charge different platform fees to sellers of different quality and save buyers’ attention to better sellers. This paper also shows that when market agglomeration is more profitable for the large platform, it may choose to ban smaller platforms, which could be socially inefficient as it may deter mediocre sellers from entering the market.