During the Industrial Revolution, an increase in agricultural machinery and decrease in manual labor drove millions to find work in fast-growing cities. In fact, between 1870 and 1910 manufacturing employment rose by more than 8 million jobs. Driving the change was mechanical automation. Inventions like steam-powered engines, the power loom, and electricity soon caused technological unemployment.
It’s a theme we’ve seen throughout history: newer, more efficient machines replacing work once done by hand. This time in the form of artificial intelligence (AI).
Reports from PwC, McKinsey, and the World Economic Forum estimate 60% of current jobs will require significant adaptation because of AI, and Goldman Sachs predicts 50% of jobs could be fully automated by 2045.
Alarmists say this will put people out of work, but similar to the Industrial Revolution, the AI revolution will just change the work. Just as people shifted from farms to factories, workers will face another generational shift. This time the AI will handle repetitive, menial tasks and free up hours of time for the average worker. Some estimates say AI is already saving workers about an hour a week, but by 2029 that is expected to skyrocket to 12 hours.
The security industry desperately needs a smarter, more effective way to operate, and AI is the answer.
AI in Security
Criminals aren’t stopping. In fact, everything from vandalism and violence to panhandling and theft is increasing, which creates unwelcome environments for customers, employees, and communities. Traditionally, companies solve this by hiring more people or adding another single point security solution to their tech stack in an attempt to deter crime. But neither of these are long-term solutions. If the company hires more people, criminals may be deterred for a time, but before long they will find a weak spot to exploit. This leads the company to up the ante again and hire still more security personnel. But security professionals are also faulty because they’re human. They experience alert fatigue, need breaks and vacations, and can miss things. And they are expensive. On the other hand, if the company brings on additional security tech, they will need more people to operate it and more people to respond to the alerts the tech generates. But the crutch of this is, either way they are relying on humans to increase deterrence. That is the most expensive and least sustainable way to increase security. It is completely unfeasible to keep hiring more personnel; it leads to a ballooning budget and a fallible workforce.
Security systems today are already good at detecting an intrusion and sending an alert. For example, they’re really effective at detecting motion and telling a human that something is moving on the site. This is so common that even simple doorbells on a house can reliably do it. However, the issue is, the security system still relies on a human to respond to the alert and initiate next steps. Reliance on human response is also complicated by alert fatigue, which is where operators are constantly notified by alerts and become desensitized, missing critical alerts and threats.
Agentic AI
This is where agentic AI can step in and make security faster, more efficient, and cheaper. Agentic AI is a type of artificial intelligence that can make decisions and take actions independently to automate multi-step processes, designed to operate more like a human employee. It adapts to different situations rather than being statically programmed. In the example of motion detection, agentic AI not only helps to classify what moved (a human, vehicle, or animal), but also determines the level of threat based on observable context and initiates the next appropriate steps--all without involving a human security guard. Actions agentic AI can take might include turning on spotlights to track the perpetrator or projecting custom messages over the speaker. It can even call out specific features of the intruder, like the color of their clothing, hats, and location on the property. If that still doesn’t scare off the intruder, the AI can make the decision to escalate further according to the specifications at that site. That might mean speaking a more imposing message, contacting security, or calling in law enforcement. Essentially, the agentic AI will deal with all of the alerts coming in, manage the ones it can, and only alert guards to problems that need human intervention. It helps remove human error and helps the guards optimize their efforts.
Traditionally, security efforts are divided into jobs to be done—detection, validation, deterrence, response, and prosecution. Prevention (detection, validation, deterrence) is worth its weight in gold over response and prosecution. Obviously, response and prosecution are important. They form a symbiotic relationship with prevention and form a complete security system. And if criminals aren’t held accountable through prosecution, deterrence efforts become ineffective because there's "nothing to fear.” However, traditional response and prosecution efforts rely on manual, time-consuming methods that can take years. It often costs more to arrest the criminal, gather evidence, compile the case, and prosecute the case than the original loss. Prosecutors are backlogged so they have to prioritize bigger cases. This means smaller thefts can go unprosecuted because it simply isn’t worth the time and money.
Agentic AI is the solution to these antiquated practices. It reduces desensitization and increases deterrence through customized messaging, spotlights that pan to detection zones, and other automated responses. Keeping deterrence levels high helps prevent crimes. If AI can help increase effective deterrence, it can help save money in prevention by optimizing human guards and prosecution, which often costs three times more than the initial loss.
Agentic AI also streamlines the evidence gathering and case building process. It removes manual processes so it is easier for investigators to build a case, saving time and resources. Reducing time in turn reduces cost, and less expensive prosecution makes holding criminals accountable easier, which validates the purpose of the security system/efforts to begin with.
The Human Advantage
AI isn’t perfect yet and is a long way from completely replacing security personnel. In fact, AI most likely will not replace all security personnel; a human will still need to strategically direct security activities playing the measure/countermeasure game that takes place between the good guys and the bad guys. AI will be really good at taking instructions from humans and performing the work needed to detect those threats and respond to them appropriately. AI excels at work that is monotonous, repetitive, and human-error prone, performing with tireless consistency day-in and day-out.
For example, as bad actors change their tactics, humans will need to be able to research trends and learn the new tactics to better detect and defeat them. AI agents can help humans do the research and find the trends, but ultimately a human has the reasoning, the creativity, and the purpose to strategically plan for a response.
Replacing Humans?
AI is not replacing humans in the security industry. Yet. In fact, I believe we are a long way from that. It will continue to improve as it learns from feedback loops from real-world data. But for now, the right balance of humans and technology can reduce cost, increase speed, and improve accuracy.
Just as the Industrial Revolution reshaped labor from farms to factories, the AI revolution is reshaping roles within industries, including within security. Agentic AI won’t lead to technological unemployment but to technological augmentation. It will handle the repetitive, high-volume tasks and reduce human error and alert fatigue, allowing human expertise to show through. Instead of replacing human security guards, agentic AI helps them be more strategic, proactive, and effective. The future of security isn’t human or AI but a balance between them that creates safer, more secure environments with a complete security ecosystem.
