What comes to mind when you hear about Artificial Intelligence (AI)? For many, it is actor Will Smith battling against the machines in the movie adaptation of Isaac Asimov’s I, Robot. In the 1950 book, the first law of robotics was “A robot may not injure a human being or, through inaction, allow a human being to come to harm.” In a way, modern adaptations of AI in the security industry are fulfilling this by protecting us from harm.
In Security industry events around the globe, AI or Machine Learning is being discussed as the solution to a growing problem in electronic security deployments. With the cost of common sensors going down, and their data generating capabilities rising, how will we process all the data to identify critical events and generate a timely response? Humans are poorly suited to monitor data from, say, hundreds of video cameras. Overload, fixation, fatigue, and boredom are some of the problems that can be eliminated by intelligent machines. The use of AI is shifting the focus from reaction and investigation to identifying threats before they act.
General Artificial Intelligence is defined by Sean Lawlor, data scientist at Genetec Inc., Montreal, Canada as “something that can understand the world it lives in, absorb input and learn topics it was not specifically designed to learn.” Going back to the 2004 movie above V.I.K.I. was the master computer that deduced from observing human behavior that we needed to be protected from ourselves. This is not generally what we find in systems that are commonly deployed in the consumer marketplace today. Most are more narrowly focused – think Alexa, whose primary talent is to interface through voice recognition.
One of the primary ways this more narrowly focused AI is being utilized today is in video analytics. Over time, detection has evolved from simple percentage of pixel change to more advanced algorithms like line crossing, vectored motion, object removed or left behind, etc. But only recently have surveillance systems begun to “learn” their environment and suggest events worthy of further investigation through the recognition of anomalous behavior. I have reviewed video that identifies persons of interest based on behavior characteristics. In this case, the intelligent sensor is a camera and the operator must still participate in the learning phase to improve the accuracy of threat detection. However, engineers are beginning to “bake in” more advanced recognition of normal human behavior and the sensors will become more accurate, more intelligent out of the box as the pool of available data grows.
Protection has always been a delicate balance of privacy vs. security and tragic events often have an effect of shifting that balance away from the privacy concerns. Many major cities and corporate or educational campuses are now blanketed with surveillance cameras. However, even as surveillance has become more pervasive, artificial intelligence can allow it to be less invasive. Often, identities of subjects are now protected by the system until anomalous, potentially dangerous behaviors are detected.
Perhaps the machines will protect us from ourselves.
At Vision Technologies, we invest in our employees to make sure that they stay informed on the latest trends and advances in technology. Our customers benefit from solutions that are researched and tested to make sure they perform effectively when deployed in the field.
Contact us for more information on how AI can improve your security experience.