In recent years, some new discoveries have already accustomed us to scenarios similar to those of the movie Minority Report. But in reality, unlike the precursors, there are computers that link the worlds of artificial intelligence and crime. From weather forecasts to self-driving cars, these technologies have been with us for at least a decade.
Recently, in this vein, scientists from the University of Chicago have developed a new algorithm capable of predicting crimes. Ishanu Chattopadhyay , a researcher who led the study, says the accuracy is 90%, all a week before they can happen.
Artificial intelligence and crime: the test case
Such applications, as already mentioned, are not new to the world. Japanese police have tried to use a similar approach ahead of the Olympics to anticipate any misdeeds before they happen. The Illinois academics used the city of Chicago as a reference for their algorithm. It divides the city into spatial pieces of about 300 meters in width, recreating a digital twin of urban environments. The new model isolates crime by observing the temporal and spatial coordinates of discrete events and detecting patterns to predict future events
Previous attempts at crime prediction often used an epidemic approach. Once described as emerging in "hotspots", the crime spread to surrounding areas. However, these tools do not take into account the complex social environment of cities and do not consider the relationship between crime and the effects of the police.
In the same words of the Chicago researchers and the Santa Fe Institute, the spatial models ignore the natural topology of the city. The transport networks respect roads, footbridges, trains and buses; those of communication respect areas with a similar socio-economic background. Their model allows for the discovery of these connections.
The results of the study
To the artificial intelligence model, Chattopadhyay and his collaborators have provided data from past records . Crimes committed in Chicago which they divided into two categories: violent (homicides, assaults and beatings) and against property (burglary, vehicle theft).
These data were used because they are more likely to be reported to the police in particular urban areas. Those in which there is a historical mistrust and lack of cooperation with the police. Such offenses are less prone to bias in law enforcement, such as drug offenses, road arrests and other minor offenses.
The team of researchers also studied the police response to crime . Analyzing the number of arrests following accidents, they compared these percentages between neighborhoods with different socioeconomic status. As a result , crimes in affluent areas have led to more arrests, while arrests in deprived neighborhoods have decreased . Crimes in slums, however, have not led to more arrests, suggesting a bias in response and law enforcement.
This implies that when researchers stress the system, testing it more thoroughly with various data, predictions about what to do next are clear. More resources are needed to arrest more people in response to crime in an affluent area. At the same time, the police will have to restrict them to areas of lower socioeconomic status
The marriage between artificial intelligence and crime still has a long way to go. Of course, there is a fundamental difference between the other predictive models and the one described in the article. The former are influenced by biases deriving from data such as political orientation, socio-economic status and the location of the neighborhood. The second is based only on the exact time and position in which the criminals act; once biases are eliminated, it is as accurate, if not more, than the others.
But Chattopadhyay himself clarifies and induces reflection. The accuracy of the tool does not mean that police departments should use it as a guide for cleaning up and probing neighborhoods for crime prevention. Instead, it should be added to a toolbox of urban policies and policing strategies for dealing with crime.
The article Artificial Intelligence and crime: in the service of the law was written on: Tech CuE | Close-up Engineering .