AI system used to assess risk of reoffending

Police in Durham are set to go live with an artificial intelligence (AI) system or Harm Assessment Risk Tool (Hart), designed to help officers decide whether or not a suspect should be kept in custody.

The system, which has been trained on five years' of offending histories data, will classify suspects at a low, medium or high risk of offending and has been tested by the force.

Data for Hart was collated from Durham police records between 2008 and 2012. The system was then tested during 2013, and the results - showing whether suspects did in fact offend or not - were monitored over the following two years.

Forecasts that a suspect was low risk turned out to be accurate 98 per cent of the time, while forecasts that they were high risk were accurate 88 per cent of the time.

This tool is designed to be more likely to classify someone as medium or high risk, in order to err on the side of caution and avoid releasing suspects who may commit a crime.

Prof Lawrence Sherman, director of the University of Cambridge's Centre for Evidence-based Policing, was involved in the tool's development and suggested that Hart could be used in various cases. These included instances such as when deciding whether to keep a suspect in custody for a few more hours; whether to release them on bail before a charge; or, after a charge has been made, whether to remand them in custody.

The scientists involved in the development of Hart also stressed that the forecasting model's output is ‘advisory’ and should not remove discretion from the police officer using it.

Please register to comment on this article