This probably is along the same lines as the predictive criminality models used in some US counties to justify giving some people higher bails and longer sentences.

The programs themselves don’t actually use any valid formulas and are based on prior regional and racial arrest histories, so the software generally would perpetuate the biases of the precincts and DAs of the area.

We’ve long established we can’t trust law enforcement with the forensic tools they have let alone give them new ones.

Get these feckin AI minority report bots out of my country.

Not sure why the link goes to another post. Here is the link to the article.

https://www.forbes.com/sites/thomasbrewster/2023/07/17/license-plate-reader-ai-criminal/

I imagine a close approximation of this AI would look something like this

fn drives_like_a_criminal(driver: &Driver) -> bool {
    driver.vehicle == Vehicles::Car || driver.vehicle == Vehicles::Truck || driver.vehicle == Vehicles::Motorcycle
}