US study indicates facial recognition is racist and sexist
Facial recognition technology has been implemented with great force in smartphones as a biometric security system. A technology that still has some deficiencies and is characterized by being somewhat racist. This technology has been implemented in China for control of society, originating the Hong Kong protests. But in the United States they have also published a study that indicates that it is racist and sexist.
This facial recognition technology seems to fail a lot depending on the gender or race of people. At least this indicates the study published by the National Institute of Technology and Standards of the United States (NIST for its acronym in English). But hey, coming from the United States is not a great novelty either.
- 5,8-inch Super Retina display (OLED) with HDR
- IP68 water and dust resistance (up to 2 meters deep for up to 30 minutes)
- 12 Mpx dual camera with dual optical image stabilization and 7 Mpx TrueDepth front camera: Portrait mode, Portrait Lighting, Depth Control and Smart HDR
- Face ID to securely authenticate and use Apple Pay
- A12 Bionic chip with next-generation Neural Engine
Easy recognition is racist and sexist according to US study
According to this study, a large part of the easy recognition algorithms misidentify the African American or Asian race. The error rate is up to 10 percent higher than for Caucasians.
But it is also that the NIST study adds that the error rate increases if we add the sex of the people. This report also highlights that identification is highly erroneous on the faces of African American and Asian women.
NIST has analyzed no less than 189 algorithms for easy recognition created by a total of 99 different developers. The study has excluded algorithms from large companies like Amazon. The reason these algorithms have not been analyzed is that they have been prevented from exploring their easy recognition algorithms.
Although revealing, it is not something we did not know in advance with facial recognition. There have been cases of smartphone unlocking by family and friends. Obviously, this technology is still a bit green and a lot of development is still missing.
Source: with the BBC