Untangling Governance Mechanisms in Open-Source AI Platforms: A Digital Nudging Approach to Enhance Large Language Model Adoption and Development

The rapid advancement of large language models (LLMs) has revolutionized artificial intelligence, with open-source platforms like Hugging Face emerging as crucial hubs for collaboration and innovation. These platforms democratize access to cutting-edge artificial intelligence (AI) technologies and foster vibrant developer ecosystems. Nevertheless, traditional platform governance approaches struggle to adapt to the unique challenges posed by LLMs, including their scale, complexity, and potential societal impact. While existing literature has explored factors influencing open-source software project success, there remains a significant gap in our understanding of the specific dynamics surrounding LLM models within these platforms. This research proposes a novel approach that leverages digital nudging theory to examine the effectiveness of various platform governance mechanisms in promoting LLM development and adoption. Our framework incorporates different types of digital nudges, including presentation nudges (leaderboards), information nudges (trial spaces), and nudges with facilitation (implementation of advanced models like Llama 3). By examining these nudges in the context of open-source AI platforms, we aim to provide insights into effective governance strategies that can foster innovation, ensure fairness, and promote responsible AI development.

Our study comprises two main components: a panel regression analysis of LLM data and a quasi-experimental design to examine the impact of implementing advanced models. The first phase of the study (Study 1) investigates how different types of digital nudges influence LLM project popularity and adoption, considering the moderating effects of model type. The second phase (Study 2) explores the impact of introducing Llama 3 on the Hugging Face platform, analyzing changes in LLM developers’ behavior output. Through this multi-faceted approach, we contribute to the literatures on digital platform governance, open-source software development, and ethical considerations in AI advancement. Study findings will inform platform designers, policymakers, and AI practitioners on effective strategies for fostering a thriving, innovative, and responsible AI development community within the rapidly evolving landscape of LLMs.