It’s not just about the percentage of false positives, it’s about how much easier it is to generate false positives.
Human face-matching accuracy is worse than software in many scenarios (“is this the man that robbed you?”) but it requires so much effort that the absolute number of false positives are low.
On the other hand, facial recognition hooked up to cctv can passively generate mountains of matches all day long for pennies.
This must just be a ban of the technology for government use, right? Can retailers still use facial recognition as part of their in-store analytics? Can Apple still sell FaceID products in Boston?
Human face-matching accuracy is worse than software in many scenarios (“is this the man that robbed you?”) but it requires so much effort that the absolute number of false positives are low.
On the other hand, facial recognition hooked up to cctv can passively generate mountains of matches all day long for pennies.