AI crime prevention and detection – 5 current applications

86 Views | Read Time 9 minutes |

Companies and cities around the world are leveraging artificial intelligence services to reduce and prevent crime and respond more quickly to crimes in progress. The idea behind these projects is that crimes are relatively affordable; this requires that law enforcement be able to sort through massive data to find useful models. Analyzing this type of data was technically impossible a few decades ago, but recent developments in machine learning are obligatory.

There is a good reason for both companies and the government to try to use AI solutions in this manner. As of 2010, the United States has spent more than $ 80 billion annually on state, local and federal spending. The total cost of law enforcement in the United States is estimated at over $ 100 billion a year. Law
enforcement and prisons make up a significant percentage of local government budgets.

Direct government spending is only a small part of how crime affects cities and individuals. Victims of crime face medical bills. Also, high crime reduces property value and forces companies to spend more on security. And criminal records can significantly reduce a person’s long-term employment prospects. University of Pennsylvania professor Aaron Chalfin reviews current research on the economic impact of crime, and most of the analyzes account for about 2% of the gross domestic product in the United States.

This article examines the applications of AI and machine learning in crime prevention. In the rest of the article below, we answer the following questions:

1. What AI crime prevention technologies are there today?
2. How are cities currently using this technology?
3. What outcomes (if any) have AI crime prevention technologies had so far?

This article is divided into two general categories: companies trying to use AI to solve crimes in different ways:

1. AI WAYS TO DETECT CRIME
2. AI WAYS TO PREVENT FUTURE CRIMES.

AI WAYS TO DETECT CRIME

Crime Detection

The city infrastructure is smarter and more connected. It provides real-time information sources for cities ranging from traditional security cameras to smart lights, which can be used to detect when crimes have occurred. With the help of AI, the collected data can be used to detect firearms and where the gunfire came from. Below, we cover a range of current applications:

Gunfire Identification – ShotSpotter

The ShotSpotter Company uses smart city infrastructure to triangulate the location of a gunshot.

According to ShotSpotter, only 20 percent of gunfire incidents involve people calling 911, and even when people report the incident, they often provide only vague or inaccurate information. They claim that their system can alert authoritie effectively at high hours with information about the type of gunfire and the exact location of up to 10 feet. Multiple sensors select the sound of a gunshot and their sensor learning algorithm triangles where the shot is made by comparing data such as the sound level and the resonance of the building when each sensor is listening.

AI security cameras – HikVision

While ShotSpotter is listening for crime, many other companies are using cameras for it. Last year, Hikvision, a leading security camera maker, announced it was using chips from China Movidius (Intel Company) to create deep neural networks that could be deployed on-board.

People announce the new camera can entirely scan for license plates on cars, execute facial recognition to search for potential offenders or missing persons, and automatically detect suspicious anomalies such as unattended bags in crowded venues. HikVision says they can now achieve 99% accuracy with their advanced visual analytics applications.

Hyquision is the number one supplier of video surveillance products and solutions in 2016, according to IHS, with 21.4% of the market share for CCTV and video surveillance equipment worldwide.

Movidius explained the benefits of building this capability directly on new cameras.Their systems have been used for many years, such as facial recognition,license plate reading, and unattended bag detection, but that video processing has traditionally taken place in a centralized center or the cloud. By processing the cameras themselves, they act Bandwidth is the only information that needs to be transmitted..After the introduction of their cameras, crime at Sea Point, South Africa, was reduced by 65%.

AI WAYS TO PREVENT FUTURE CRIMES.

The goal of any society is not just to catch criminals, but to prevent crime from happening, and in the examples below, we will explore how this can be achieved with artificial intelligence.

Ting of future crime spots – Predpol

PredPol is a company that uses big data and machine learning to predict when and where crimes occur. They state that when and where new crimes occur, they can analyze existing data on past crimes. Currently, their system is present in many American cities, including Los Angeles, and is the earliest recipient.

Their algorithm revolves around the observation that certain crime types are clustered in time and space. They claim that they can predict where future crimes will take place by using historical data and examining where recent crimes have taken place.
For example, burglary rashes in one area may be associated with more burglaries in surrounding areas shortly. They call this method real-time epidemic-type aftershock sequence crime forecasting. Their system highlights possible hotspots on the map, and police should consider patrolling more.

A victory highlighted by Tacoma, a Washington firm, saw a 22 percent drop in residential burglaries as soon as the system was adopted. Tacoma began using Predpole in 2013, and in 2015 the burglaries fell.

Since crime is a complex issue for many reasons, it is very difficult to separate the effectiveness of any one tool. However, according to a study by researchers at Predpole, police patrolling based on real-time epidemic-type aftershock sequence crime forecasting (what Predpole uses) reduces crime by 7.4%.

 Cloudwalk

Chinese facial recognition company Cloudwalk Technology is trying to determine if a person has committed a crime before. The facial recognition and gait analysis technology helps us use advanced AI to track and track people.

The system will find out if there are any suspicious changes in their behavior or unusual movements. For example, if a person seems to be running behind a certain area, they may be pickpocketing or casing that area for a future offense. It tracks the person over time.

Pretrial Release and Parole – Hart, and COMPAS

After being convicted,most people will be released until they are prosecuted. Determining who should be released as a pretrial or what a person’s bail is in the past is now mainly done by judges using their best judgment. Within minutes, the judges had to try to determine if there was a risk of flight, a serious threat to society, or harm to the witness if released. It is an imperfect system open to bias.

The city of Durham in the United Kingdom is using AI to improve the existing system of determining the release of the accused. The Harm Assessment Risk Tool (HART) program provides five years’ worth of criminal data. Hart uses that data to predict whether a person is a low, medium or high risk.

The city has been testing the system since 2013 and compared its expectations with real-world results. Hart’s estimates that a person is at low risk are accurate 98 percent of the time, and the city’s estimate of a person at high risk is 88 percent accurate.

The idea is to advise Heart officials that suspects may be involved in another crime. The jurisdictions in the United States have been using more basic risk assessment algorithms for over a decade to make decisions about pretrial release and whether or not a person should receive parole. Correctional offender management profiling is the most popular for alternative sanctions (COMPAS) from Equivalent, which is used in all Wisconsin and many other places. The 2012 analysis of the New York Division of Criminal Justice Services found COMPAS to be “effective in the recidivism scale and achieving satisfactory assessment
accuracy.”

COMPAS fires were
recently blamed after a ProPublica investigation. The analysis of the media
company suggested that the system may indirectly have a strong racial bias.
ProPublica’s coverage of COMPAS is critical

This report raises the question of whether better AI can ultimately produce more accurate predictions or reinforce existing problems. Any system is based on real-world data, but if real-world data is produced by partisan police officers, it will make AI biased.

Concluding Thoughts and Future Outlook

The ability of AI to allow governments to collect, track and analyze data for policing purposes raises some serious questions about privacy, and machine learning can create a feedback loop that reinforces organizational bias. This article is not devoted to these important issues, but the AI  Now Institute at New York University is a research center dedicated to understanding the social implications of artificial intelligence that can provide more detail on these issues.

Despite civil liberties concerns, they have so far not spread AI technology in surveillance and crime assessment. According to IHS, 245 million professionally installed video surveillance cameras were operating in 2014, and the number of security cameras in North America more than doubled from 2012 to 2016. More data is being provided to security and law enforcement agencies; it is only natural that they will want to invest in more and more AI tools to keep up with this ever-increasing data flow.

The use of AI and machine learning to detect crime through sound or cameras currently exist, which has proven to be work and is expected to continue to expand. The use of AI in predicting crime or the likelihood of a person committing a crime is promising, but it is not yet known. Proving that it works for politicians is the biggest challenge. When a system is designed to stop something from happening, it is difficult to prove a disadvantage. Companies directly involved in providing governments with AI tools to monitor the area or to assess crime can benefit from a positive feedback loop. Improvements in crime prevention technology will increase the overall cost of this technology.

Above all, the low crime rate has broad social benefits for a society and a real political advantage for local elected officials who are responsible for the budget. Liberal mayors such as Bill de Blasio in New York City and conservative mayors such as Rudy Giuliani have heavily claimed that their re-election campaign will reduce crime during their tenure.

Most of these technologies, developed primarily for government clients, have spillover benefits for private companies. Private companies are also using the same AI security cameras used by the government to protect their assets. The technology that can be used to assess crime or to catch suspicious behavior automatically can help companies decide where to avoid risk or new locations.