> The UK’s National Physical Laboratory’s data shows the system being tested and used by UK police to search their databases returns the correct identity in 99% of cases. This accuracy level is achieved by balancing high true identification rates with low false positive rates.
1% is accurate when you are imagining a one-off identification, like a police line-up. But when you are conducting millions of scans a day that means you are triggering tens of thousands of incorrect identifications per day (assuming equal type i/ii errors). Many of those are going to be identifying an innocent person as a wanted suspect, or pinning people as being near a crime who weren't actually anywhere near, or identifying a lawful immigrant as someone that overstayed a visa etc etc.
Is that why ICE got two photos of immigrants for for a US citizen, that they took in despite having proof of citizenship, only to find out his proof was right and the facial recognition was twice wrong?
> The UK’s National Physical Laboratory’s data shows the system being tested and used by UK police to search their databases returns the correct identity in 99% of cases. This accuracy level is achieved by balancing high true identification rates with low false positive rates.
1% is accurate when you are imagining a one-off identification, like a police line-up. But when you are conducting millions of scans a day that means you are triggering tens of thousands of incorrect identifications per day (assuming equal type i/ii errors). Many of those are going to be identifying an innocent person as a wanted suspect, or pinning people as being near a crime who weren't actually anywhere near, or identifying a lawful immigrant as someone that overstayed a visa etc etc.