[chadgpt3]

I don't agree. Stereotypes that are wrong but get reinforced by selective reporting are bad. Let's say a person believes that (to pick an absurd example) all Scottish people are rapists. Every time he reads a headline "Scottish man rapes woman" his belief will strengthen, but when he reads "Irish man rapes woman" it won't weaken. And of course there are no headlines saying "Scottish man does not rape woman".

Ubiquitous surveillance and publication of crimes has the potential to industrialize this process. If there are more rape headlines going around, then of course there will be more where the perpetrator is Scottish and the ratchet advances more quickly.

Especially if you use machine learning which easily reinforces biases. If you train it on lots of Scottish rape cases it'll learn a lower threshold for detecting Scottish people as rapists and then your bias has translated into a real statistical difference in detected crimes and even a difference in arrests and convictions.