6 Ways AI Is Making the World a More Secure Place
By Natalie Daelemans
Home Security Cameras
There are lots of companies out there selling home security systems designed to keep your house safe from intruders. However, the cameras these systems use can be pretty limited. They detect motion and pick up sounds, but then send alerts to you—the human—to sort through notifications that could be potential threats or could just be a friendly squirrel playing on your porch. With AI technology integration, security cameras now have the potential to become a much more powerful tool to protect your home.
Ooma and BuddyGuard are two companies with AI powered security cameras on the market. These cameras are able to learn from what’s going on around them. Facial recognition technology allows the cameras to distinguish between family and strangers, and they’re smart enough to know the difference between suspicious sounds, and normal neighborhood noises (such as your neighbor’s yappy dog or the sprinklers that go off every day at four).
Assisting Police Officers
A lot of police work involves sorting through clues (or data), recognizing patterns, and making connections. That takes up a lot of time. What are computers great at doing? Sorting through data, recognizing patterns, and making connections. And they can do it in a matter of seconds. It therefore isn’t surprising that AI is being integrated into several aspects of police work.
One example is an AI called VALCRI, a system which is able to analyze data from crime scenes and previous police cases to determine possible explanations for crime investigations. With a single click it can sort through mounds of data that would take a person hours to comb through, and then present possible connections and explanations. Crime analysts then tell the computer how relevant those connections are, helping the AI become smarter and more effective over time. Police are still involved throughout the whole process, but their time is spent doing the actual thinking while the computer takes over searching for clues and presenting plausible crime scene narratives.
Another AI assisting police is this one, which is being used to help identify possible serial killers. Investigating serial killers has always been an area of police work that has been particularly challenging. It sometimes takes a while before officials even realize that a series of homicides are connected and that the same murderer is responsible for multiple deaths. That’s why artificial intelligence is especially helpful. Because it can sort through information from hundreds of cases at once, AI is able to identify possible connections between cases and alert authorities of possible serial killers.
In China, police are being given a lot more power to keep tabs on criminals. Surveillance cameras across the country now have facial recognition technology allowing authorities to track the movement of suspicious individiuals, foreigners, and locals alike. It may be a bit creepy and there are concerns about the government having too much control over its people, but the database of faces has come in handy in apprehending quite a few criminals. The cameras work in conjunction with special smart glasses that some police now wear. The glasses (kind of like Google glass) allow police to scan faces in a crowd and identify whether suspicious people match any of the faces they have on profile in their system.
Keeping Police in Check
While the police are there to keep us safe, there’s also been a lot of turmoil recently surrounding the subject of police shootings. You might be glad to know that AI not only has the potential to assist police, but also to keep them accountable. A group of researchers has trained an AI to scan newspaper articles from around the country to compile a list of people killed by police officers. The system still has a long way to go, but it indicates the direction AI is headed and the kind of public information we will someday be able to monitor.
Lots of time and money can be wasted when seeking out potential money laundering schemes. Human-led investigations can often lead nowhere. With AI seeking out suspicious activity instead, resources aren’t wasted and more crime can be detected. Since 2012, this AI has identified enough money laundering to result in over $1.92 billion in fines to parties involved in these criminal activities within the U.S.
Manually scanning through surveillance video is another activity that takes up a ton of time. Most cameras are on 24/7, but there are usually only a couple of minutes of footage per day that hold any interesting information. To make the process easier, this AI, called Ella, allows you to search through video the same way you might search Google. Type in a key word, such as “white truck,” and the deep-learning cloud-based search engine scans all the footage to find any clips that have a white truck. Once it’s hooked up to a camera’s feed, Ella begins scanning and tagging items it sees, and becomes more aware over time.
Helping Lawyers Spot Liars
In a courtroom, telling a lie is sometimes, literally, the difference between life and death. That’s a lot of pressure on judges, lawyers, and jurors, who have to make calls based on what they’re told by witnesses. But advances in machine learning could prove helpful. This AI was shown video of mock court cases and asked to determine whether the witnesses were lying or not. To detect lies, the AI monitored things like voice patterns and micro-expressions on the face of those speaking. These little clues are too small or imperceptible for people to notice, but for a computer, they are important pieces of data that lead to pretty impressive results: a score of 90 percent accuracy. The technology isn’t ready to be implemented in actual cases yet, but could eventually play an important role in doling out justice.
More Secure Password Alternatives
So much of our lives and personal information is stored in the digital world, and protected by nothing more than the 8+ characters we use as a password. Hackers are becoming more and more adept at cracking your password and unlocking private information. UnifyID is a machine learning system that uses other, more secure ways to identify you. It lives as an app on your phone, and as you go about your day-to-day activities, it collects information about your behaviors and habits that are distinctly you. Things like the way you walk and the cadence of you typing. It builds up a personal profile of data that is intrinsic to you to make personal authentication more accurate and secure.
About Natalie: Natalie is a communications major and English minor at Brigham Young University. She is a student account manager at the BYU AdLab and plans to continue working in advertising after her graduation in April. This blog post was written as part of the Applied Digital Communications Practicum.