039

neta:studio_ai:news_show:039

A Cambridge Forum paper on archival provenance argues that AI data practice can learn from archival ideas about creators, custody and context, while archives can also learn from modern dataset documentation. It proposes treating data as archives rather than context-free material.