Artificial intelligence can predict crime a week in advance. It has a success rate of 90%
According to results published in the journal Nature Human Behaviour, the model performed comparatively well in all eight U.S. cities. It was this that contributed to its creation.
These were Atlanta, Austin, Detroit, Los Angeles, Philadelphia, Portland, San Francisco and Chicago.
An artificial intelligence tool is able to predict crime based on learning time and geographic patterns.
These will be evaluated on the basis of violent and property crimes.
Data scientists at the University of Chicago have trained a computer model from public data from eight major U.S. cities.
Artificial Intelligence and Crime Prediction
However, it turned out that even this model has its drawbacks. It does not take into account systemic prejudices in police judgement, nor its relationship to crime and society.
It is sometimes more than complicated.
A kind of bias against groups or races has already been shown in the past. This could largely be repeated in this case.
However, the researchers say their model should primarily serve to inform current policing strategies.
These, in turn, could compensate for the disparity in policing.
Socio-economically disadvantaged areas could thus receive more police attention than has been the case so far.
The ratio between poor and rich neighborhoods would be equalized.