These days, the last thing any company wants is to be put on blast by social media or a watchdog group with a good reputation and powerful following (like the ACLU).
Unfortunately for Amazon, the ACLU conducted an experiment using the company's new facial scanning software, Rekognition, to show just how dangerous it could be for law enforcement to be using facial scanning tech. In short, when the ACLU ran the faces of the members of the United States legislature through the software, it returned false positives for 28 members of Congress.
While the concept behind using facial recognition software seems great, there's certainly a problem when it comes to accuracy. For consumer purposes, such as unlocking that fancy iPhone X or Samsung, facial recognition is a viable alternative to other biometrics, like a fingerprint.
But when it comes to policing, there's still a long way to go before the tech is ready for deployment. Without some serious legal requirements/restriction placed on the use of this tech, as the ACLU experiment shows, there's certainly a very high risk that individuals will be misidentified using facial recognition tech, which could lead to severe consequences.
The ACLU is not alone, as recently even Microsoft joined in, in demanding that Congress act to regulate the use of facial recognition.
Facing the Numbers
The ACLU test also showed what many have been warning against from the start: AI discriminates against minorities. The ACLU test's numbers showed that "nearly 40%" of the false positives occurred to the members of Congress that are people of color.
The test, apart from trying to exert more pressure on Amazon to not sell facial recognition software or services to law enforcement, was meant to be a wake up call to the members of congress that facial recognition tech could impact them personally as well.