Just today I was preparing some slides for a course that makes the student participate and show some topics we're interested in regarding social and ethical issues in compsci
I googled a bit and I found a hn comment that talked about an article which explains what you say
Basically, they estimate drug use in the population, and they draw a map, and black and white people more or less have the same habits in drugs use. They draw a map of the past arrests by the police for drug crimes and they're skewed towards the black neighborhood. There's a bit more about it since the article dealt with predictive policing and Ai but the gist is what you're saying, and this causes a self reinforcing feedback for the police (which in turn arrests more black people and the loop goes on)
Yeah, that's quite a strong assumption. Basically, they're assuming that the local drug use is similar to the national one. While not perfect, they provide some other cases in which these data were deemed more trustworthy than police reports. at least to me it is quite crystalline that a system relying on past police data to predict future crime will be inherently biased by the fact that the police data itself is biased towards certain demographics
I googled a bit and I found a hn comment that talked about an article which explains what you say
Basically, they estimate drug use in the population, and they draw a map, and black and white people more or less have the same habits in drugs use. They draw a map of the past arrests by the police for drug crimes and they're skewed towards the black neighborhood. There's a bit more about it since the article dealt with predictive policing and Ai but the gist is what you're saying, and this causes a self reinforcing feedback for the police (which in turn arrests more black people and the loop goes on)
https://rss.onlinelibrary.wiley.com/doi/full/10.1111/j.1740-...