The algorithm is performing its designed task-- to give people a chance of lesser punishment if the algorithm decides that they probably won’t be detrimental to society. However not everyone benefits from this algorithm, it is believed that more people benefit from it than are harmed, making it ethical to a utilitarian.
I believe that these algorithms can be ethical by some deontological theories, but the US criminal justice system itself follows teleological theory. In the deontological approach, the algorithms take inputs of what someone has done in their life and develops an output to return depending on the rules in the algorithm. Example algorithm:
Inputs: [X-killed four people, X-assaulted someone] Rules:[Murder is bad, Assault is bad] Consequence: X-probably is not safe for the