Question from "Hi Phi Nation Predictive Policing Podcast"

Question

When it comes to predictive policing, the softwares such as PredPol or CSOP claim to be unbiased due to the computer automated decisions. However, the "hot spots" that the police must patrol can still contain a large amount of bias. That being said, my question is: "Does predictive policing lead to a higher amount of false accusations and more biased interactions with people in an area due to the fact that the software predicted that there would be crime in that area?"

Answer 1

I can relate to your argument because I experienced this when I book my AirBnb rooms in the past. I usually look up the "gang map" on Google and I am feeling less secure whenever my AirBnb apartment near such area. The same goes with the police as they become highly alarmed with the "hot spots" in their consideration. Hence, combining with the rising trend of false accusations based on biases, such as races and origins, I do think that the people in the high surveillance areas will have a higher chance of being falsely accused.

Answer 2

I definitely think there is a danger of many false accusations because of these hot spots, but I think there is a point where we can minimize these false accusations and still have some benefit to predicting real crimes. However, how many of the true crimes will we miss if we try to actively minimize the false accusations? Will these missed crimes diminish the benefit of this program to the point where the benefits are overshadowed by the loss of privacy?