"Even if face recognition can match a face with 99% accuracy, the sheer amount of faces available in police databases makes false positives inevitable. (The 1% error rate means that, if 10,000 people who are not wanted by the police undergo face recognition, 100 will be flagged as wanted)."
The author is not being fair.
Usually with technologies like these, developers will not aim for an accuracy of 99%, but a precision of 99% (that is what we do with Congestion Charging in London, where the ramifications of FPs are much lower [I work in Transport for London]). That means that for 10,000,000 people undergoing facial recognition, only 100 may be flagged as wanted, and only 1 will be a false positive. If we were talking about 99% accuracy, that doesn't necessarily mean what the author claims either. Accuracy is (TP + TN)/(P + N), meaning that the decrease in 1% from 100% can be in any of:- a lowering of TP or TN; or an increase of FP or FN. There is no reason to think that the 1% will all mean false positives.
"Even if face recognition can match a face with 99% accuracy, the sheer amount of faces available in police databases makes false positives inevitable. (The 1% error rate means that, if 10,000 people who are not wanted by the police undergo face recognition, 100 will be flagged as wanted)."
The author is not being fair.
Usually with technologies like these, developers will not aim for an accuracy of 99%, but a precision of 99% (that is what we do with Congestion Charging in London, where the ramifications of FPs are much lower [I work in Transport for London]). That means that for 10,000,000 people undergoing facial recognition, only 100 may be flagged as wanted, and only 1 will be a false positive. If we were talking about 99% accuracy, that doesn't necessarily mean what the author claims either. Accuracy is (TP + TN)/(P + N), meaning that the decrease in 1% from 100% can be in any of:- a lowering of TP or TN; or an increase of FP or FN. There is no reason to think that the 1% will all mean false positives.