Main content

Why police have turned to algorithms in the fight against crime

14 February 2019

Durham Constabulary made headlines in 2017 when it announced it would introduce an artificial intelligence system to help determine whether suspects brought into custody would be likely to re-offend.

It was a controversial move, prompting the question of whether algorithms should be involved in police decision-making at all.

Professor Lawrence Sherman of Cambridge University helped Durham police introduce the algorithm; he explained on Brainwaves why the scheme was introduced.

How are algorithms used to prevent crime?

Prof Lawrence Sherman explains how algorithms can forecast patterns of offence.

“We don’t need to help police forces decide which cases are most serious. They know that. What they don’t know – and want help with – is to decide which of these hundreds of thousands of not-very-serious offences have been charged to people who turned out to be very serious offenders.”

Professor Sherman explained that the algorithm works by exploring patterns across a total history of prior arrests, including: the number of their offences; the type and severity of offence; whether they’ve been brought to a custody suite before; and the age at which they were first brought to a custody suite for a serious offence.

“In tens of thousands of cases, we see very different patterns,” claimed Professor Sherman.

“Patterns that sometimes experienced officers can look at the rap sheet and think they understand. But the big data adds much more precision and much more accuracy to making a forecast from the exact pattern that each offender has in their prior criminal record.”

How AI helps Chicago police deal with violent crime

Chicago's crime-predicting software

Chicago battling violence with crime-predicting tech

Latest features from ³ÉÈË¿ìÊÖ Scotland