So who is going to analyze the automated systems for bias and correct it? If it is impossible to identify and fix unfair bias in humans, presumably it is impossible to fix in the human overseers of this system. This is one area where technology will not save us.
It could be a matter of scale - being able to improve the bias in a human doesn't mean you can do so for all 900,000 sworn law enforcement officers in the US (or any other country).
On the other hand, you can update 900,000 computer processes in a relatively short time.
The 900k computer processes are unlikely to refuse to follow the updated protocol because it's immoral/fascist/illiberal/undemocratic/whatever too. Cuts both ways it seems.
