The UK government launched a passport checker that used facial recognition to match photos to people's faces, despite knowing that the system was more difficult for ethnic minorities to use.
In response to a freedom of information request submitted by the New Scientist, the Home Office said: “User research was carried out with a wide range of ethnic groups and did identify that people with very light or very dark skin found it difficult to provide an acceptable passport photograph. However; the overall performance was judged sufficient to deploy.”
The facial recognition system, which has been in place since 2016, has difficulties correctly recognising the faces of people of colour. It mistakenly identified a black coach's lips as an open mouth, as reported in the Evening Standard, and failed to identify a black woman's open eyes and mouth.
Today, I simply wanted to renew my passport online. After numerous attempts and changing my clothes several times, this example illustrates why I regularly present on Artificial Intelligence/Machine Learning bias, equality, diversity and inclusion. #passport pic.twitter.com/sEsmdTcR1L — Cat Hallam (@CatHallam1) April 6, 2019
Speaking to the New Scientist, a spokesperson for the UK’s Equality and Human Rights Commission said that “a person’s race should not be a barrier to using technology for essential public services. We are disappointed that the government is proceeding with the implementation of this technology despite evidence that it is more difficult for some people to use it based on the colour of their skin.”
Facial recognition software has repeatedly demonstrated a bias towards people of colour. Amazon's Rekognition software was found to be nearly perfect when identifying the gender of men, but would only recognise women with dark skin tones 31 percent of the time.
Competing technologies from Microsoft, IBM, and Megvii were also challenged when asked to identify pictures of dark skinned people in comparison to light skinned people.
In the UK, use of the technology has come under criticism for the way it has been implemented in cities. Citizens have been concerned over the lack of consent given for the cameras to be used, and fines that have been given for people avoiding the systems.
A number of US cities - San Francisco, Oakland, California, and Somerville, Massachusetts - have banned the technology because of concerns over its misuse.