Mayra Russo

Reflections on the unseen and underappreciated aspects of all the (data) labor that goes into making data intensive systems work!

The machine learning (ML) breakthroughs we are experiencing as a society, and commonly talked about under the Artificial Intelligence (AI) signifier, are possible today in part due to technical advances, but also due to the existence of the coveted "data" used to fuel these systems, and the inordinate amount of human labor that contributes to the elaboration and preparation of datasets. End-to-end data life cycles typically fail to recognize not only the large amounts of time spent on data tasks (i.e. acquiring, cleaning, transforming data) performed by researchers, data scientists, and other professionals, but also all the previous human intervention and labor that is needed for the elaboration of datasets to come to fruition in the first place (i.e. capturing, interpreting, annotating, and curating data) by an invisible workforce made up of administrative staff, office clerks, local experts, and gig workers, among many others.