AI ‘predicts crime with 90 per cent accuracy’
Image credit: Foto 34913566 © Juliengrondin | Dreamstime.com
An artificial intelligence algorithm that has been able to accurately predict crime in Chicago also revealed existing biases in police responses.
Scientists from the University of Chicago have developed a new algorithm that they claim can predict future crimes one week in advance with about 90 per cent accuracy.
The AI, which has been taught to identify patterns in time and geographic locations using public data on violent and property crimes in the city of Chicago from 2014 to the end of 2016, was able to accurately predict crime levels several weeks in advance. The model was also trained and tested on data for seven other major US cities, with a similar level of performance.
However, the AI's conclusions also suggested the existence of bias in police responses across the city, associated with race.
According to the research, published in the journal Nature Human Behavior, crime in wealthier areas of Chicago resulted in more arrests than in poor neighbourhoods.
“What we’re seeing is that when you stress the system, it requires more resources to arrest more people in response to crime in a wealthy area and draws police resources away from lower socioeconomic status areas,” said Ishanu Chattopadhyay, senior author of the study.
The tool was tested and validated by focusing on two broad categories of reported events: violent crimes (homicides, assaults, and batteries) and property crimes (burglaries, thefts, and motor vehicle thefts). Those were the crimes most likely to be reported to police in urban areas.
Previous efforts at crime prediction have often used an epidemic or seismic approach, where crime is depicted as emerging in 'hotspots' that spread to surrounding areas. These tools often overlook the city environment as well as the relationship between crime and the effects of police enforcement.
“Spatial models ignore the natural topology of the city,” said sociologist and co-author James Evans. “Transportation networks respect streets, walkways, train and bus lines. Communication networks respect areas of similar socio-economic backgrounds. Our model enables discovery of these connections.”
The new model divides the city into spatial tiles roughly 1,000 feet (300 metres) across and predicts crime within these areas instead of relying on traditional neighbourhoods or political boundaries, which are also subject to bias.
Previous efforts to use AI tools to predict crime have been controversial because of the risk of bias. When the Chicago Police Department trialled an algorithm that created a list of people deemed most at risk of being involved in a shooting, it was revealed that 56 per cent of Black men in the city aged between 20 to 29 appeared on it.
Chattopadhyay conceded that the data used by his model will also be biased, but says that efforts have been taken to reduce the effect, and that the AI doesn’t identify suspects, only potential sites of crime. The team has also released the data and algorithm into the public sphere, so that other researchers can check its reasoning and conclusions.
“We created a digital twin of urban environments," said Chattopadhyay. "If you feed it data from what happened in the past, it will tell you what's going to happen in future. It's not magical, there are limitations, but we validated it and it works really well.”
Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.