The American Civil Liberties Union (ACLU) has published a fiercely critical test of Amazon Rekognition, Amazon’s new facial recognition system.
The ACLU compared photos of American politicians with Amazon Rekognition’s existing database of mug shots. It falsely matched 28 politicians, primarily black and Latino people, with those mug shots.
It seems to be another instance of AI and risk assessment software displaying prejudice against non-white people. However, Amazon Web Services representative Nina Lindsey told the New York Times that the technology would be used in a different way by law enforcement.
It is worth noting that in real world scenarios, Amazon Rekognition is almost exclusively used to help narrow the field and allow humans to expeditiously review and consider options using their judgment.
Amazon Rekognition can identify whether an object is a person and also identify aspects of a face, such as expression. It aims to identify specific people in images and can detect up to 100 faces in a crowd. It has already faced criticism for its alleged racial prejudice. Amazon employees have also spoken out against it, writing an open letter urging Amazon not to let US immigration force ICE use the technology.