There is an even deeper problem.
That these AIs are designed for terrorizing citizens and not stopping crime.
"Show me the man, and i will show you the crime." said by the leader of the "crime prevention" that resulted in many people disappearing off the streets.
The way these "AIs" (they are not artificial intelligence) currently work is to arrest someone three times by mistaking them for someone they were looking for. Yes, the police broke someone's constitutional rights three times, because a computer said so. Facial recognition completely fails.
Facial recognition is racist. No, really.
It can more easily tell someone's race than it can someone's sex.
And to hide that fact, there has been a lot of kludging done on the software (making it less effective)
Further, the database which this "AI" is pulling information out of us is woefully incomplete and VERY biased. The people in it are mostly good citizens. Because they interact with the DMV and have had their children finger-printed, etc.
Even further, the database of crimes is mostly "failure to pay ticket".
Who broke into your home and stole your money and jewels? No one knows. They aren't in a database. They don't have bank accounts. They don't have phone plans in their own name. The police don't come to the scene of the crime and take fingerprints; they barely take down a police report.
The crime we would like to stop isn't even in the database. And if it is not in the database, it can never come up as the result of a query.
And before you say, "But what about all the criminals who have been busted before, they are in the database". And yep, they are there, but they rarely have paper trails. Many are wanted for missing bail or probation and are not found for years, if ever.
So, this "AI" is only for terrorizing honest citizens.