Certification systems for machine learning: Lessons from sustainability

Concerns around machine learning’s societal impacts have led to proposals certify some systems. While prominent governance efforts date center networking standards bodies such as the Institute of Electrical and Electronics Engineers (IEEE), we argue that learning certification should build on structures from sustainability domain. Policy challenges share significant structural similarities, including difficult observe credence properties, data collection characteristics or carbon emissions model training, value chain concerns, core-periphery inequalities, networks labor, fragmented modular creation. networking-style typically draw their adoption enforcement functional needs conform enable network participation, learning, despite its digital nature, does not benefit this dynamic. We therefore apply research systems in sustainability, particularly commodities, generate lessons across both areas, informing emerging EU’s AI Act.