Researchers publish a multilayer dataset of the AI supply chain
Original title: Mapping the AI Supply Chain: Dependencies, Vulnerabilities, and Market Concentration in the AI Ecosystem
The public dataset tracks how models, data, chips, capital, and talent move between organizations and flags concentration and chokepoints.
A paper maps organizations in the AI supply chain and the relationships among models, data, compute, capital, talent, and other assets. The authors release a dated, tagged multilayer dataset and use it to examine central actors, network concentration, communities, and chokepoints. They say public records omit private deals and that search design shapes coverage. The data are on Hugging Face and can be explored at aisupplychains.org.
Unconfirmed: The authors say the dataset is necessarily incomplete because public documents omit private deals.
- The pipeline annotates actors and relationships with dates, descriptions, types, and tags.
- Analyses cover central actors, concentration, communities, and chokepoints.
- Authors say private deals are missing and coverage depends on search design.
- Data are at huggingface.co/datasets/sarahcen/ai-supply-chain and aisupplychains.org.
From the original article
1 min read at arxiv.orgAI systems are built across many organizations that contribute models, data, compute, capital, talent, and other assets. While the layers of the AI stack are relatively well understood, the relationships between actors and components across layers remains sparsely documented.
By Sarah H. Cen · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.