E-Handbook: Computer vision AI looks beyond the narrow into the mainstream Article 4 of 4

alphaspirit - Fotolia

AI in policing is controversial but delivers results

The drive toward more predictive policing is only growing stronger as computer vision and other AI tools are put to use. But not everyone is happy with the trend.

Scanning video footage for anomalies; augmenting surveillance and crowd control capabilities; transcribing audio; deploying facial recognition technologies. These are just a few of the ways law enforcement agencies are using AI technologies to boost their effectiveness. The use of AI in policing now allows law enforcement agencies to do more than they ever could before to keep the peace, identify and prevent threats and speed up the time it takes to catch criminals and solve crimes.

AI helps detect threats

When there are large gatherings such as parades, festivals or sporting events, law enforcement agencies depend heavily on technology to help keep a watchful eye over the crowd and detect threats. Police departments now use AI-enhanced image and video technology to help with crowd control and surveillance of certain heavily populated areas. AI-enabled computer vision systems assist officers in public spaces such as sports venues, train stations and airports, providing constant surveillance.

Facial recognition technologies are also increasingly being used, especially in China, to help identify criminals in large crowds. Notably, Chinese authorities have picked out individual criminals among stadiums of tens of thousands of people through facial recognition technology. One Chinese city, Guiyang, has a network of over 10,000 cameras throughout the city to help police identify and arrest suspects. Big brother is watching you, indeed!

Chinese company Hikvision is creating cameras with built-in deep neural networks. This enables these cameras to detect suspicious anomalies in crowded areas such as unattended bags or an out of place car. Sometimes it's just not possible to have boots on the ground everywhere law enforcement may need to be. These AI-enabled cameras help to keep an eye out for suspicious activity all hours of the day.

In the United States, facial recognition technology is used to help identify and find missing persons. According to NCIC Missing Person and Unidentified Person Statistics, in 2017 there were over 88,000 people missing in the U.S. alone. Facial recognition technologies are increasingly being used in public places such as stadiums or airports to help identity missing individuals and suspected human trafficking operations. In addition, AI in policing helps categorize, classify and prioritize inbound tips from the public. Many agencies are understaffed and have difficulty keeping up with their current workload and the volume of tips. AI is being put to good use in this area. In early 2017, Intel partnered with the National Center for Missing and Exploited Children (NCMEC) to help comb through the millions of tips they receive each year, automating much of their analysts' manual work, and substantially increasing the number of valid tips that law enforcement could process.

Law enforcement departments and intelligence agencies are also using AI technology to monitor social media to help spot and identify radicalized individuals or illicit activity such as drug dealing or gang activity. With increased use of various social media platforms such as Facebook, Instagram, Twitter or Tinder, criminals have platforms to identity potential clients or victims, and also communicate their crimes or intentions. Through the use of machine learning and AI-based content monitoring tools, agencies can increase the ability for their agents to keep an eye on social media activity. This application of AI in policing flags suspicious accounts and conversations that could potentially require further investigation by human agents.

Increased use of robots expands capacity

Robots have long been used to help police departments with tasks that are deemed dangerous to humans, such as detonating bombs or gathering intelligence in hostage, robbery or active shooting situations. Now these robots are equipped with AI technologies to provide more sophisticated capabilities. These robotic assistants are not only able to detect, but also deactivate, explosive devices. In the not too distant future, an AI-powered robot could be employed in a hostage or active shooter situation, autonomously finding and identifying suspects, and potentially acting further to neutralize the threat.

In 2017 Dubai police their robotic officer, with the department stating that they want to have robots make up a quarter of their police force by 2030. The robot can speak six different languages, is equipped with a computer touch-screen to allow people to report crimes, and cameras designed to read facial expressions and transmit live images back to the operations room where human counterparts are monitoring. These robots, currently located mainly at tourist spots around the city, are meant to be assistive and not replace human police officers.

AI assists with non-violent crime

In addition to violent crimes, there are other types of crime that police must investigate such as stolen and counterfeit goods, financial crimes and petty theft. These crimes impact quality of life, but many resource-strapped departments don't have the manpower or expertise to efficiently investigate and prosecute these crimes.

High-end goods such as purses, sunglasses and clothes are frequently counterfeit and identifying these fake products can be difficult. With the help of AI in policing, it's becoming easier for law enforcement to spot and identify fakes. Startup company Entrupy has developed a device that can take microscopic photographs of multiple areas on an item and run it though algorithms that analyze various materials such as leather, wood or metal. The company claims more than a 98% accuracy rate in detecting counterfeits, and provides users with a certificate of authenticity if the item is real.

AI algorithms are also able to help investigations into financial crimes such as expense report fraud or stolen credit cards. Banks can train AI algorithms with historical data, and coupled with transaction monitoring systems, they can be used to verify or flag transactions that look out of the ordinary.

Potential misuse of AI in policing

Police departments are able to take advantage of the benefits of AI technology to fight and solve crime but also face pushback from communities that wish to preserve privacy and security in the process. Many of these technologies are still , so laws and regulations have not caught up with their use. Some people argue that these technologies can be misused and potentially violate rights or disproportionately target minorities and people of color. Police departments in cities such as Orlando, Fla., and Detroit are piloting facial recognition technologies within their departments.

Because it's in a pilot phase, policies and guidelines are not yet in place to help keep these departments in check. Facial recognition systems have been known to be less accurate identifying people with darker skin tones. Many commercial facial-analysis services currently used by police departments were found to be far less accurate in identifying black women than white men according to a 2016 study conducted by Georgetown.

Will law enforcement's use of AI make us safer as a society, or do some of these technologies sacrifice freedom and privacy for security and convenience? Regardless of the outcome of this debate, it is clear that AI is making an indelible mark on the future of policing and law enforcement.

Dig Deeper on AI business strategies

Business Analytics
Data Management