We have talked about the potential use of AI in detecting crime before. We discussed how police departments in Great Britain were working on AI programs to identify criminals before they commit any crime. All moral ambiguity aside, there is a lot of potential here and it seems it is not limited to Great Britain. The NYPD have begun using algorithms to spot patterns in criminal activity which allows them to more quickly identify possible suspects. We are going to talk about what exactly they are trying to accomplish long term.


This software was recently developed by the NYPD goes by the name of Patternizr. This software allows crime analysts stationed in each of the department’s 77 precincts to compare robberies, larcenies and thefts to hundreds of thousands of crimes logged in the NYPD’s database, transforming their hunt for crime patterns with the click of a button. This method has proven to be much faster than the old method, which involved analysts sifting through reports and trying hard to remember key details about various crimes. Then they would have to decide whether they fit into a pattern or not.

This tool also has proven to be more comprehensive, too, allowing analysts to spot patterns across the entire city of New York as opposed to just in their precinct. Patternizr picks up key details from crimes logged across all the precincts which gives each individual office access to key insights that normally would have been rather difficult to put together.

The software, which had begun development back in 2016 has so far helped NYPD detectives solve hundreds of crimes since it was rolled out in the beginning of March 2019. This is very significant progress in only a short amount of time proving the usefulness of their tool, Patternizr. The two developers who created the algorithm were already on staff which ensured the cost to develop it was minimal.


The department disclosed its use of this technology only this month encouraging other departments to develop it as well citing how effective it has already proven to be in only a month’s time. When asked what their goals are regarding the use of this software the NYPD said that the goal is to improve safety. They also added that the more easily they can identify patterns in crimes the more quickly they can apprehend the perpetrators which could lead to lower overall crime rates.

They say they were inspired by a team out of New York University who was working on this concept but never produced a workable model. The NYPD wanted to take things a step further to make their jobs more efficient. Obviously safety is everyone’s primary concern when dealing with criminals but it is safe to say that their goals extend far beyond public safety. The NYPD wants to streamline the entire process using AI to save time and money. Regardless of whether they care more about that or about lowering crime rates this technology will end up accomplishing both.


The team who developed the algorithm that powers Patternizr used a variety of methods to train it and to ensure accuracy in pattern recognition. They began with providing the algorithm 10 years of data and patterns on past crimes that humans had painstakingly put together by themselves. Early on when testing the algorithm it was able to re create these patterns of the past from the given data about a third of the time and returned parts of each pattern 80 percent of the time. The accuracy of the algorithm improved quickly and today it re creates proper patterns nearly 100 percent of the time.


The old method of doing things was too slow and too limited in scope. Given the success the NYPD has had so far it is safe to say that other departments will be quick to adopt these kinds of programs as well. The closer and closer we get to super AI the more developments like these will become a reality. The war between criminals and law enforcement rages on, and AI will play a huge role on both sides.

Leave a Reply

Your email address will not be published. Required fields are marked *