A Cartography of Open Collaboration in Open Source AI: Mapping Practices, Motivations, and Governance in 14 Open Large Language Model Projects
A new study published on arXiv examines the collaborative practices within 14 open-source large language model projects. Researchers interviewed developers to understand how these models are initiated, organized, and governed throughout their lifecycle, from development to reuse. The study found that collaboration spans various domains like models, data, and software, with motivations ranging from democratizing AI access to promoting open science. Findings suggest that openness in AI is not a fixed trait but an outcome of how collaboration is structured across different stages and contexts. AI
IMPACT Provides insights into the dynamics of open-source LLM development, potentially guiding future collaborative efforts and governance models.