In 2014, Duke researchers set up cameras to record thousands of individuals walking on West Campus, with the objective of collecting a data set that could enhance facial recognition technology.
They published this data in 2016, and the data set—known as DukeMTMC—has since been cited and used in at least 175 additional papers, some of which have been tied to the Chinese government’s oppression of the Uyghur minority. The Duke webpage with the data set and details of the project was recently taken down.
The letter of apology from one of the researchers is available here.
This story is a chilling account of the issues with data collection and surveillance in our modern society. We are constantly under surveillance, and if it were not for the work of the Duke Chronicle, it remains to be seen if this story would have come to light. We need to consider the many of stories like this (and worse) that never come out.
Research ethics questions aside…this story is an example of the challenges of collecting data and sharing it openly online. Your data can be shared widely online, and used for a variety of purposes. In an age of deepfakes, and a need to train algorithms…this data is a goldmine.