The most devastating tool for fighting crime is not increasing the penalties or making jail more miserable. No, the most effective way to stop crime is by drastically increasing the chances of getting caught and correctly sentenced.
> Asian and African American people were up to 100 times more likely to be misidentified than white men, depending on the particular algorithm and type of search. Native Americans had the highest false-positive rate of all ethnicities, according to the study, which found that systems varied widely in their accuracy.
One. Hundred. Times. The camera may say it recorded John Smith breaking into a house, but it’s incredibly likely that it was actually Ron Jones. Who’s the jury going to believe, though: John Smith who was at home with his girlfriend at the time, or the hugely expensive video system that the city just bought?
Facial recognition doesn’t work. It’s bullshit tech, and we should stop using it until we make it deliver on its promises AND decide how to deal with its ramifications.
[1] https://sitn.hms.harvard.edu/flash/2020/racial-discriminatio...
[2] https://www.wired.com/story/best-algorithms-struggle-recogni...
[3] https://www.washingtonpost.com/technology/2019/12/19/federal...
Furthermore, an accurate system is not the same as a morally just system. A ubiquitous system which recognizes people with 100% accuracy could be used for evil if it ever fell into the wrong hands. Such powerful systems should never be created in the first place, the juice isn't worth the squeeze.
The racial bias issue is still important for now, but it's fast becoming irrelevant. We should be asking ourselves where our priorities lie even if bias weren't a concern.
A video completely indistinguishable from the video presented by the prosecution. Not sure where we go from there.
No we look at the primary suspect's digital footprints and analyze metadata, look for other nearby video feeds, etc. We try to see if additional evidence is corroborating.