Hacker News new | past | comments | ask | show | jobs | submit login
Police Misuse of Facial Recognition – Three Wrongful Arrests and Counting (skynettoday.com)
51 points by andreyk on Aug 12, 2021 | hide | past | favorite | 16 comments



Where's the human in the loop?? I have no problem with the computer pointing out possible bad guys, but that's only *possible* suspect!

Same problem as considering a field drug test to mean anything more than you need to do a real test.


See 'Facial Recognition Should be Used with Care' - the police detectives should have been the 'human in the loop' in these cases, but appear to have done their job poorly in all these cases.


That's a novel use for AI: excuse for slacking. Joke aside, that's also how most of the "driverless" accidents happened - they expected the human would jump in but they were doing what any human would have done - slacking. Because AI is supposed to help us, right.


They were likely slacking before AI came along. AI just helps them boost their numbers without actually doing anything.


That's probably because police are incredibly undereducated for the power they wield.


Hey HN! I am an editor on this piece, which I think highlights a trend not enough people are aware of. Would definitely welcome feedback / any knowledge wrt these kinds of news stories you think is relevant.


Tawana Petty https://tawanapetty.org/ is a Detroit based activist who has been at the fore front of the movement addressing algorithmic bias in policing.

In a country based on the precept that all Black people are the “bad guy” until proven otherwise, any and all facial surveillance should be ended. References for that statement? How about Angela Davis: Davis, Angela. Abolition Democracy: Beyond Prison, Torture, and Empire. New York: Seven Stories Press, 2005.


Thanks for letting us know!


You should do a piece on how the racial biases used to determine the "correct" color balance for film grew into the algorithms for facial recognition and result in the terrible accuracy for non-white non-males.

https://sitn.hms.harvard.edu/flash/2020/racial-discriminatio...


What is missing is figures about how many cases facial recognition helped solve.

Without figures, it's difficult to tell if it brings more troubles than benefits.

It really sucks for those wrongly convicted, but it may cause more harm not to use this technology.

An immediate way to alleviate those injustices would be to instruct police officers that statistical models do not spit absolute truths. Great tools are not a substitute for either thinking nor humanity.


True - this article presents a nice overview with more details on that https://www.nytimes.com/2020/01/12/technology/facial-recogni...


Along the same lines, I'd be interested in seeing what is the difference in incorrect arrests in communities with facial recognition and without. Its terrible that anyone is falsely accused, but if the use of the technology decreases incorrect arrests thats a valuable trade-off.


I'd like to see how many cases facial recognition helped solve compared to just throwing dice.

As a society our decision used to bw that you are not convicted unless proven guilty. This principle is not a minor thing and maybe you choose the wrong thing to optimise here?


How will it cause more harm not to use this technology? It can't possibly cause more harm to not just it on the future than not using it in the past.


> An immediate way to alleviate those injustices would be to

pay restitution to the victims of abuse at the hands of the State.


I wonder how it compares to normal police detective work. In theory if a human is there to verify things, these types of mistakes wouldn't happen.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: