Tech Trends: AI - The Good, the Bad, and Perhaps the Ugly

May 10, 2023
As more uses of AI and robotics emerge, should we consider it friend or foe?

This article originally appeared in the May 2023 issue of Security Business magazine. When sharing, don’t forget to mention Security Business magazine on LinkedIn and @SecBusinessMag on Twitter.

The total destruction of mankind…or maybe not.

The advancement of technology over the past three decades is astounding. Just a short 20 years ago, the vast majority of video surveillance systems were still using VHS tapes, multiplexers, and analog cameras with resolutions that were just a fraction of todays most basic of imaging technologies. Patrol, response, and investigations were exclusively done by manpower.

Today, we are seeing a massive influx of true Artificial Intelligence – albeit mostly basic video-based analytics – into these traditional security technologies.

Basic AI in Practice for Security

We are now able to see the emergence of AI’s most elementary abilities surfacing in the security industry. An example is Coresight.ai, whose technology is able to provide facial identification of individuals using standard video surveillance cameras – even when the face of the subject is partially obscured. The technology provides the ability to match a person of interest on a watch list to individuals captured on live and recorded video.

Consider Zeroeyes and Omnilert, whose AI technologies use standard video surveillance cameras to detect a person in possession of a gun, prompting the system to trigger an action or alert prior to the subject carrying the gun making entry to a facility or shooting the weapon.

Companies like Viisights bring behavioral analytics that are able to detect a wide variety of behaviors like fighting, crowd formation, accidents, slip-and-falls, throwing objects, or other aggressive behaviors.

The vast amount of currently installed security technology is wholly reactive, meaning that an incident must take place prior to a mitigative action being taken. These technologies make security systems and departments proactive in a way we have never seen. AI can greatly speed the detection and response to many incident types, and also trigger on precursors that happen before an actual incident occurs.

As for AI and robotics, by now you have likely seen Spot the robot dog from Boston Dynamics, and perhaps some of you have seen Atlas, the most dynamic humanoid robot in the world from the same company. There have been plenty of similar robots roaming tradeshow floors the past couple of years.These robots now have human-like agility and the ability to analyze their environment. They are equipped with an array of sensors that give the robotic platform detailed information on its movement, surroundings, and position, as well as process incredible amounts of data in milliseconds. The robot can calculate thousands of probable scenarios, weighing what ones have the most likelihood of occurring, then almost instantly react to the situation, faster than humans are capable.

Imagine the Possibilities

You can easily extrapolate the possibilities. Consider the integration and convergence of just these few AI technologies and robotics mentioned: A person with a gun exits their vehicle at a school or bank, the AI determines the presence of a gun and also determines that the person in possession of the gun is on the facility’s watch list (placed on the list perhaps because they are a terminated employee or a student identified as high risk), this triggers an automated lockdown, pre-scripted announcements, law enforcement notifications, starts live video streaming to responding agencies, and automatically tracks the location and actions of the subject. Then a security robot, powered with this real-time information, moves into place to intercept the subject as they attempt to gain entry into the facility.

The robot can analyze threats, verbal exchanges, micro expressions and movements of the subject against thousands of like scenarios or previously documented incidents to make a determination of the subject’s probable actions before the subject even takes them. Then with faster-than-human reaction time and agility, engage the subject with less-than-lethal force to disarm and subdue the subject.

While we are not quite there yet, it is certainly on the horizon. In April, the New York Police Department announced the use of Spot the robot dog and other robots to start patrolling Times Square.

Playing Devil’s Advocate

To say this article is far-reaching would be a false statement; so why would I choose to open with such a negative statement like “the total destruction of mankind?”

Let’s entertain another scenario, an adversarial one. We know that all technology eventually becomes more commonly understood and can be used for both good and bad. Currently, the development of AI and robotics technology has little to no regulation or oversight, providing few limits to what this technology can be programmed to do, or better yet, to not do.

Elon Musk is quoted as saying AI is “quite a dangerous technology” and that AI has the potential to destroy civilization. What happens when an adversary – criminal enterprise, enemy state, or the AI itself – obtains and programs AI to do their bidding?

Imagine an adversary who develops an integrated AI robot with advanced sensory technology and physical agility, the ability to analyze and process petaflops of information to render an action hundreds of times faster than any human’s ability. The outcome is potentially unstoppable. It is a frightful view into a “Terminator” movie-type threat that security and law enforcement teams are not remotely ready or equipped to respond to.

Avoiding the Ugly

We should not be surprised that the security industry is embracing the newest emerging technologies such as AI and robotics. As illustrated, they are an obvious force multiplier and bring new efficiencies to improve the security function.

However, the bad is there, just bubbling under the surface.

In the end, the security industry should embrace a better approach to develop and use this technology with some firm ground rules – to hopefully harness the good and avoid the bad and potentially ugly.

Paul F. Benne is a 35-year veteran in the protective services industry. He is President of Sentinel Consulting LLC, a security consulting and design firm in based in New York City. Connect with him at www.linkedin.com/in/paulbenne or visit www.sentinelconsulting.us.