Scraped Data, Hidden Faces
Installation Scraped Data, Hidden Faces explores the role of repetetive, manual human labour in the context of machine learning. The work is inspired by the specific case from the end of 2021. At that time, OpenAI Company hired dozens of workers from Kenya, Uganda and India through an outsourcing firm outside Nairobi to detect violent, racist or otherwise inappropriate content in the machine learning datasets and manually remove it. These workers were paid less than two dollars an hour for this activity, and many of the developed psychological problems as a result of the dehumanizing work and repeated confrontation with drastic or otherwise unpleasant material. At the same time, however, the resulting machine learning outputs lack sufficient representation of similar populations.The dataset capturing the visual representation of workers‘ labour as collected by the internet search engine is permeated with black boxes with plain writing making the undignified labour of „workers of the machine learning era“ more „visible“. The piece was created during the symposium RESISTANCE: Decolonising Europe’s Digital Transformation in one of the old villas on the Greek island of Lesbos.
(the collages above the fireplace are the work of Polish filmmaker Tomasz Melnik)
installation from printed photographs, 174cm x 141cm
