Publication Type
Master Thesis
Version
publishedVersion
Publication Date
6-2026
Abstract
This paper studies how generative AI reshapes decentralized production by examining the release of ChatGPT as a shock to open-source software development. Using a panel of non-AI GitHub repositories and exploiting cross-language variation in ChatGPT’s early coding performance, I compare repositories written in languages more compatible with ChatGPT to those written in less compatible languages. I find that treated repositories experience a significant increase in contribution concentration after the shock. This increase is driven primarily by the exit of peripheral contributors, but also by a residual reallocation of work toward top contributors. The reallocation effect is concentrated in repositories without organizational affiliation, consistent with less formal governance structures allowing core contributors to absorb tasks more quickly. I then examine why peripheral participation declined. The evidence suggests two complementary channels. First, generative AI increased the creation of new high-quality repositories, especially among organizationally affiliated projects, suggesting that it lowered the cost of initiating new projects and reallocated effort across the open-source ecosystem. Second, peripheral contributors’ willingness to participate did not decline substantially; instead, their contributions became less likely to be incorporated into existing projects. Despite these changes, I find no evidence that generative AI increased output within existing projects; if anything, output declined, particularly among organizationally affiliated repositories. The findings suggest that generative AI expands the extensive margin of open-source production while contracting and concentrating production within existing projects.
Keywords
Generative AI, Organizational Structure, Open-source Environment, Software Development
Degree Awarded
Master of Philosophy in Econ
Discipline
Industrial Organization | Labor Economics
Supervisor(s)
XU, Jianhuan
First Page
1
Last Page
65
Publisher
Singapore Management University
City or Country
Singapore
Citation
LI, Zhaozhen.
The impact of generative AI on open-source projects: Evidence from GitHub. (2026). 1-65.
Available at: https://ink.library.smu.edu.sg/etd_coll/919
Copyright Owner and License
Author
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 4.0 International License.