Wenfei Xu, David Levinson, Michael J Smart, and Nebiyou Yonas Tilahun summarizing their paper, “The racial composition of road users, traffic citations, and police stops” in The Conversation (a new-to-me “fact-based” nonprofit news organization):

Our research, published in June 2024, used data on the racial composition of drivers on every street in Chicago. We then compared who is driving on roads with who is being ticketed by the city’s speed cameras and who is being stopped by the Chicago police.

Our findings show that when speed cameras are doing the ticketing, the proportion of tickets issued to Black and white drivers aligns closely with their respective share of roadway users. With human enforcement, in contrast, police officers stop Black drivers at a rate that far outstrips their presence on the road.

For instance, on roads where half of drivers are Black, Black drivers receive approximately 54% of automated camera citations. However, they make up about 70% of police stops.

On roadways where half of the drivers are white, white drivers account for around half of automated citations – and less than 20% of police stops.

“Grimm”, snarking on Mastodon:

COP UNIONS RIGHT NOW: This is definitely an issue. We need to train AI to be racist.

Brutal, but fair.

The article goes well beyond the paper, with examples of the consequences of Driving while Black (DWB) and ways to improve policing and enact police reform.

There’s no denying DWB is real; I’ve experienced it myself multiple times. Removing human bias from policing and similar decisions can be beneficial, but I’m not a fan of increasing the “surveillance state,” especially because—as we’ve seen with almost every AI or automated system—our human bias is often baked into the system.

For example, they note in their paper that:

the location of the cameras themselves may not be [race-independent]

because

cameras are not placed in a race-neutral way

and acknowledges that

police stops do not occur on random streets but are selective of specific streets.

This all suggests that cameras may be deployed—and policing may occur—more in Black and Latino neighborhoods than in white ones.

The cameras may be race-neutral, but the people placing them are not.

(Via Paul Cantrell.)