Researchers at the University of Chicago have developed an algorithm that predicts crimes within a radius of 300 meters a week in advance with 90% accuracy. The study was published in the journal Nature.

The scientists say the system learned patterns from publicly available data on violence and property crimes. Such incidents are less prone to police bias, the researchers say.

The model isolates crime by looking at the temporal and spatial coordinates of individual events and discovering patterns to predict future events.

The algorithm divides the city into “spatial tiles” about 300 meters wide and predicts crime in those areas.

Previous models have relied more heavily on traditional neighborhoods or political boundaries that are subject to bias, the scientists said.

According to the study’s lead author, Ishanu Chattopadhyaya, the tool’s high accuracy does not mean it should be used to guide law enforcement policy. He added that police departments should not use it to preemptively seize neighborhoods to prevent crime.

“This is not magic; There are limitations, but we tested it and it works very well,” Chattopadhyay said.

Instead, the scientist argues, the algorithm should be added to the toolkit of urban policy and police strategies to combat crime:

“We have created a digital twin of the urban environment. If you tell him what happened in the past, he will tell you the future.”

The research team also examined police response to crime by analyzing the number of arrests after the events and comparing rates across different areas.

They found that higher crime rates in wealthier neighborhoods led to more arrests. Scientists say the situation is reversing in disadvantaged areas, which may indicate an imbalance in police response.

The tool was tested using historical information from Chicago. The model also performed well with data from seven other US cities: Atlanta, Austin, Detroit, Los Angeles, Philadelphia, Portland, and San Francisco.

Recall that in August 2021 the Pentagon tested a decision-making system that could predict events several days in advance.

In September, DeepMind developed a rain forecasting algorithm.

Follow ForkLog AI on TikTok!