Artificial intelligence is usually talked about as software. But every model depends on a physical system: minerals and specialised equipment, chips and data centres, electricity and water. Those dependencies also affect the places and people around them.
The problem is that nobody has the full view. Regulators, companies, researchers, journalists and campaigners each hold pieces of it, often collected for different reasons and described in different ways.
In sectors like clothing, timber and minerals, structured data is already used to trace products to their origins, assess risks and enforce rules. We explored how far existing standards and approaches could take us towards a clearer picture of AI supply chains.
Our experiments looked at four questions:
- Describing the system: Can the UN Transparency Protocol describe an AI model and its supply chain?
- Connecting information: How easily can information from fourteen AI model cards be compared and connected?
- Understanding who needs what: Who needs information about AI supply chains, and where do they stand on disclosing it?
- Following relationships: What can public data tell us about lobbying and funding for AI supply chain transparency?
The experiments showed that there is already a strong base of standards and approaches to build from, but important gaps remain.
Seeing how the system fits together depends on information being described consistently, connected across organisations, and shared in ways that match what different people actually need to do.
The experiments showed where existing standards could already help us describe and connect parts of the AI supply chain, and where gaps in information, shared definitions and identifiers still make that difficult.
More of our work
Open Source Tools and Documentation
Technical guides and references for open, adaptable data tools.
Sustainability Canvas
A conversation guide for data initiatives working towards long-term impact.
Artificial Intelligence in Development & Humanitarian Systems
Use cases, lessons, and IATI’s role in responsible adoption.