John Faithfull๐๐ช๐บ๐ด๓ ง๓ ข๓ ณ๓ ฃ๓ ด๓ ฟ๐งก โ๐ผโ๐ฟ
Prof. Emily M. Bender(she/her)It was obvious from the outset that predictive policing was a scam (not to mention a vehicle for accelerating overpolicing), but it's really valuable to have this careful reporting of just bad it is:
Misuse Case@emilymbender The algorithms for predictive policing were always fed by crime and arrest data generated fromโฆsurveillance and enforcement primarily targeted at poor and/or non-white people! So guess what those algorithms are going to tell cops to do more of? Bias in, bias out.
As I said in the post you are directly responding to: "It was obvious from the outset that predictive policing was a scam (not to mention a vehicle for accelerating overpolicing)"
This site remains 'splainy AF.
Steven Capobianco@emilymbender First time I heard about this weeks back, I thought โIโve seen enough Minority Report to know where this is headed.โ
Big Kumera Energy@emilymbender sigh, and yet they just keep signing up, including letting the 'ai' decide if they use weapons
Colin Rowat@emilymbender I'm guessing that the underlying methodology was based on the seismological approach in https://doi.org/10.1198/jasa.2011.ap09546 . If so, has there been a proper analysis of why this isn't appropriate? (1) biased data (mentioned in comments), (2) small numbers don't allow averages to help, (3) something else?
@emilymbender I've just seen _Predpol_ mentioned in Granville & Granville's _Prime Suspects_ (p.216).
๐๐ท๐ญ๐๐ฃ๐ฒ๐ฎ๐ญ๐๐ฎ ๐ค@emilymbender It doesn't even work in the movies.
J.H.NoyesI would say it's so bad those that sell it should be charged with fraud.