Algorithm based policing and the risk of disproportionate police activity

Presentation at the session "Predicting Problems? Big data, algorithms and underlying issues: Models for successful policing"

The use of prediction in the world of policing and security is not necessarily a new phenomenon, particularly with the advent of geo space analysis and the information and data now available to support law enforcement. It is used to predict crime trends, behaviour patterns relating to certain types of criminality and even to recognition of individuals via face imaging, utilising algorithms, which search and analyse historical data. Indeed,
in the UK one police force is developing an algorithm, which utilises historical data from within the criminal justice system to decide whether an individual should be granted bail or not following an arrest but before charge. However, an overreliance on the use of such an approach may not just be flawed but may increase the unwarranted attention of the police to those marginalised groups who may already receive disproportionate
police activity. This presentation examines the use of such algorithms and big data sets and discusses some of the inherent problems that appear to lie within this process and which practitioners and policy makers alike should be aware of.