Team AA group of law enforcement officers and data analysts recommend using machine learning algorithms to set police patrol routes, i.e., which times and which neighbourhoods to patrol and how often. They claim that using machine learning will allow for more efficient use of officers' time and equipment, reduce administrative staff, and lead to safer neighbourhoods. They also claim that a machine learning route planning solution will eliminate bias among the police force since no human is making the decision on which neighbourhood might get more or less police presence.
Team BLocal community activists and the Canadian Civil Liberties Association argue against the use of machine learning in setting police patrol routes. They argue that machine learning algorithms are inscrutable mechanisms that cannot justify or explain their decisions, and therefore should not be used to delegate patrolling decisions. They also argue that because machine learning requires data to build a model upon which to plan the patrols, some factors from the data will be fed into the machine learning algorithm that will influence the patrol decisions the model makes. Unfortunately, the existing data is already biased by current and past patrol, arrest, prosecution, and sentencing decisions, and therefore should not be used.