Minority Report soon? New AI tech to predict crimes weeks ahead with 90% accuracy

By On Jul 1, 2022

Photo : iStock

KEY HIGHLIGHTS

Accuracy of crime rate in any particular area reported at 90 percent.
Tech will only predict area and amount of crime in a particular time period in that area and not suspects or nature of crime.

New Delhi: A technology, based on artificial intelligence is claimed to predict locality and rate of crimes in an area with a staggering accuracy of 90 percent by its researchers, giving out strong vibes of Steven Spielberg‘s classic film, Tom Cruise starrer, Minority Report.

Researchers at the University of Chicago have created an artificial intelligence (AI) model that analysed crime data of two years from Chicago, Illinois, from 2014 to 2016. It then gave out the crime data for a couple of weeks after the analysed data. As per the end result, it predicted the numbers and areas of the crime with a 90 percent accuracy.

The areas were divided into stretches of 300 metres across each at least a week in advance. After the success in Chicago, the AI model was also tested for other major US cities and it gave out results with similar accuracy and effectiveness.

Related News

Robo helpers ‘Temi’ to assist passengers at Bengaluru airport

Wearable trackers with AI can spot COVID-19 symptoms before they appear: Study

Previous such efforts to create a “Minority Report” like machines had ended up with racial bias. The researchers involved in developing of this AI based tech said that this tech will also expose those previous efforts.

According to the report on newscientist.com, the Chicago Police department had tried predicting crimes based on a similar algorithm while keeping the identities of the data involved under covers. However, when the identities were revealed, 56 percent of the suspects were revealed to be black people aged between 20 to 29 years.

Researchers were reported saying that this tech might also fall prey to bias but the fact that it only predicts the rate of crime in a particular area and not the particular suspects, this bias might be significantly reduced.