Ghosts in the Machine

Main Article Content

Nina Toft Djanegara


The National Institute of Standards and Technology's (NIST’s) Facial Recognition Vendor Test is considered the gold standard for assessing the performance of facial recognition models. Since the test series began in 1994, it has been regarded as an objective benchmark for corporations and academic research groups to compare their models against one another. To evaluate the accuracy of facial recognition models, NIST draws upon a database of millions of photos of US visa applicants and border crossers. Without their knowledge or consent, immigrants’ photos have become raw material for the refinement of facial recognition software. This essay discusses the afterlife of these images, photos of immigrants and non-citizens that have been repurposed for a function that goes far beyond their original intent.

Article Details