Guarding Ground Zero: A look at the amazing and terrifying artificial intelligence that will protect the new World Trade Center

The site of the worst disaster in modern American history is being born anew. Right this second, construction crews are working tirelessly to rise five new skyscrapers, the National September 11 Memorial & Museum, the World Trade Center Transportation Hub, and a performing arts center, on the 16-acre World Trade Center site. Upon completion next year, 1 World Trade Center will soar 1,776 feet into the sky, making it America’s tallest building, and one of the tallest skyscrapers in the world.

Unfortunately, the rebuilding of the World Trade Center also reestablishes it as Target #1 for would-be terrorist. To combat this gruesome threat, the Port Authority of New York and New Jersey, and other WTC shareholders, are pulling all the stops to build a state-of-the-art surveillance system that aims to prevent a repeat of the September 11 attacks. According to The New York Post, the multi-million-dollar security system will feature a vast array of “military-grade” surveillance technology, include face recognition systems, retina scanners, and fully-automated, highly “intelligent” cameras that use artificial intelligence software to automatically detect any “unusual movements,” and report such irregularities to the NYPD and other authorities.

This last bit caught my eye: Intelligent cameras that learn on their own, and automatically alert the authorities when anything is out of place? That sounds like something out of dystopian science fiction. I had to take a closer look.

While the WTC security system will employ technology from a wide array of companies, including Diebold and SightLogix, the bulk of the system is being outfitted by Houston, Texas-based Behavioral Recognitions Systems, or BRS Labs. BRS Labs makes a product called AiSight (“eye sight”), which uses artificial neuro network (ANN) technology, allowing it to “autonomously identify abnormal behavior within the field of view of a surveillance camera in real time,” according to the BRS Labs website. Anytime something unusual happens within view of an AiSight-powered camera system, an alert is sent to the appropriate authorities.

For example, say a court house requires visitors to pass through a metal detector when they enter the lobby of the building. AiSight will learn that, normally, people who pass in front of the surveillance camera that’s trained on that portion of the room go through the metal detector. But if someone tries to instead go around the metal detector, rather than through it, AiSight will pick up on the abnormal motion, and send an alert to the building’s security team, in real time.

Like other artificial intelligence systems, AiSight uses computer learning to build “memories,” or “Hypocepts,” as BRS Labs calls them, based on what it “sees” through the cameras to which it’s connected. These memories become more and more refined over time, as the AiSight system gathers and analyzes more and more information about the environments it watches.

“In this industry, there are far too many cameras being deployed, and not enough human monitors to evaluate the data being captured by those cameras,” said BRS Labs President John Franzzini in an interview with Government Security News. The problem doesn’t end there. “Even if a person could be dedicated to watch every camera, studies show that a human’s effectiveness dwindles after only 20 minutes on the job,” says BRS Labs in a promo video for AiSight. “Within an hour, there is no value to the dedication simply because the human brain cannot focus on the task at hand.”

Even though video analytics systems have been around for a decade or so, Franzzini insists that these systems are fatally flawed due to the massive amount of upkeep and customization required to make them work properly. And even then, these expensive systems pale in comparison to AiSight.

The problem with other video analytics systems, says BRS Labs, is that they are based around rule-based logic systems, which require programmers to define every possible scenario that the system may or may not need to detect. This necessitates massive amounts of man-hours to accomplish, and is limited both by the imagination of those programming the system, and the technical capabilities of these systems to detect actual threats, rather than false alarms.

AiSight is different. Instead of a rule-based logic system, AiSight is “reason-based,” which allows it to automatically set up the parameters for its functionality, and improve them as it processes more and more data — no humans necessary.

While other video analytics software requires a programmer to tweak the system anytime something in a camera’s field of view changes, AiSight’s flexibility lets it adjust on the fly. So if, for instance, a camera is trained on a parking lot, legacy video analytics systems would have to be custom tailored to the specific lighting conditions and movement present in that space. But what if it snows? Other systems would have to be reworked entirely. AiSight adjusts automatically, and continues to chug along unfazed.

The benefits of AiSight are not simply that it learns on its own, but that it could prove more effective at actually preventing crime. Standard surveillance systems have been shown to do little to actually stop people from breaking the law. According to a 2005 study from the University of Leicester in the UK (home to the world’s largest public surveillance network, CCTV), the presence of a surveillance cameras did not actually reduce crime. Yes, the footage can be used by law enforcement as evidence after a crime is committed. But that’s hardly acceptable at a site as vulnerable to catastrophic terrorist attacks as the WTC.

Because AiSight automatically detects suspicious behavior, like a car parked in a strange location, a person wandering around in circles, or someone snooping where they shouldn’t, and immediate contacts the authorities, it’s possible for AiSight-powered systems to stop terrorism or other dangerous crime before it actually happens. Combine this with the WTC’s facial recognition and iris scanners (which can identify if someone is in the federal database), and the chances of another disaster occuring there become greatly lessened.

You may have notice that, so far, I’ve spoken positively about BRS Labs and AiSight; from my research, it seems like a vast improvement over previous surveillance software systems. But that’s just the tech geek side of me talking. The suspicious, civil liberties advocate side is thinking, “Creeping Jesus! This is the beginning of a terrifying Orwellian police state!” Which very well may be true. Not that much can be done to stop it — nor would most people would want to, anyway.

According to an ABC/Washington Post poll from 2007 (via Popular Mechanics), 71 percent of Americans supported an increased use of video surveillance. And given the history of the WTC, I can only imagine that only the most advanced, effective security system at that site will do. Moreover, US law firmly supports the use of such surveillance in public spaces.

Modern “search and seizure” laws under the Fourth Amendment are defined by the 1967 US Supreme Court case, Katz vs. United States 389 US 347. In its decision, the Court declared that, “What a person knowingly exposes to the public, even in his own home or office, is not a subject of Fourth Amendment protection, but what he seeks to preserve as private, even in an area accessible to the public, may be constitutionally protected. Generally, a person walking along a public sidewalk or standing in a public park cannot reasonably expect that his activity will be immune from the public eye or from observation by the police.”

In a more recent case, United States vs. Sherman, 990 F. 2d 1265 (9th Cir. 1993), the Court of Appeals for the Ninth Circuit found that video surveillance on public streets does not violate the Fourth Amendment’s prohibition of “unreasonable searches and seizures.”

Despite the legal support for video surveillance, public advocacy groups warn that the law is not yet sufficiently advanced to deal with modern surveillance issues, especially with concern to digital images, facial recognition, or any of the other technologies the WTC system will employ.

From a 2006 New York Civil Liberties Union “Special Report” on video surveillance in New York City (pdf):

There is only a limited recognition in the law that there are some places into which a surveillance camera is not allowed to intrude. And there are virtually no rules that prohibit police or private entities from archiving, selling or freely transmitting images captured by a video surveillance camera. The courts have yet to address the fundamental privacy and associational rights implicated by the phenomenon of widespread video surveillance.

In short, the world of video surveillance exists in a kind of lawless Wild West, where the governments and corporations involved can do what they like, at least until someone comes in and stops them. We see this type of situation with the Internet all the time. Something tells me this issue is only going to become more complicated and problematic in the world offline, as technology like AiSight becomes more widespread.

The balance between public safety and individual privacy is a tough thing to achieve, as anyone who’s tried to fly on an airplane in the post-9/11 world knows all too well. On the one hand, I completely support protecting the World Trade Center in every way possible. I was in NYC on the bright Tuesday morning of September 11, 2001. I saw the towers fall with my own eyes, inhaled the ghostly gray smoke that rose from the rubble with my lungs, and shared in the collective shutter of sadness and disbelief that rippled through all 8 million of my fellow New Yorkers on that fateful day. There’s no question: That disaster can never be allowed to happen again.

And yet, while I possess a visceral understanding of the kind of threats that face the new World Trade Center, I also fear the ubiquitous watchful eye of the state baring down on all of us like a subway train. I believe it is important that the new technology that keeps us more safe does not also make us less free. Alas, technology like AiSight will, at least for the time being, likely do both.

So when the World Trade Center opens to the public next year, remember a few things when you go to visit: When you’re taking pictures of the America’s new tallest building, remember that it’s taking your picture right back. And, above all, make sure to not do anything weird, or you might end up in an interrogation room.

[Images: WTC rendering via Silverstein Properties/WTC.com, eye via Semisatch/Shutterstock]

This article was originally posted on Digital Trends

More from Digital Trends

NASA rovers carried World Trade Center aluminum to Mars

Opinion: Will Windows on ARM challenge the iPad where Android has failed?

Opinion: Is returning to the moon really such a crackpot idea?

Beneath the Surface: How Microsoft is missing the boat on tabletop computing