It seems that this software predicts higher levels of crimes in areas that have had higher levels of crimes in the past. That's not bias--it's realism. And it would be inefficient to send police in equal numbers to all neighborhoods regardless of crime level.

Fair point but the article does mention that the model trains on reported crimes, and also suggests that Black homes tend to report crimes that White homes may not. So the training data has a bias, which is reflected in predictions.

Also, more police presence means more crimes are tracked. That doesn't mean they aren't happening in other areas, but without equal measurement, it may seem like areas have different counts.

So, worth considering if the training data is biased, and adjusting for that in the build and validation phases.