A challenge experiment was just conducted in Guiyang, China to see how long it would take to locate and capture a BBC reporter who was mixed in at random with the general population of 3.5 million people. After the police snapped his picture and fed it into the massive facial recognition database, which contains the facial image of every Guiyang resident, the reporter took off. They gave him a head start as he mingled in on the crowded city streets; then a request for his apprehension was given to the computer equipped with the latest AI security software.
The AI software combed through millions of images from tens of thousands of cameras in Guiyang, all transmitted back to the master computer, hunting for the reporter. The results were shocking.
It took a total of 7 minutes before police physically apprehended him. Assuming it took a few minutes for police to assemble and walk/run to his location, the actual identification could have taken as little as two minutes. Yes, just two minutes to find a single subject in a city of 3.5 million!