Key Takeaways from NENA 2024 — Part 3

Previous blogs (part 1 and part 2) shared what was learned by MCP subject-matter experts during the recent National Emergency Number Association (NENA) conference and trade show. This blog wraps up our coverage.

Excitement about artificial intelligence explodes — At last year’s NENA conference, there was considerable curiosity regarding how artificial intelligence might be leveraged by public-safety entities, including 911 centers, to improve emergency response, but understanding of this rapidly evolving technology still was at a nascent stage. This year, AI generated the loudest buzz by a wide margin, both on the trade show floor and in educational sessions. A four-hour workshop devoted to AI was held on the final day, in which MCP subject-matter experts Chris Kelly and Jason Malloy participated as panelists.

Understanding still is evolving. For instance, it’s vitally important to grasp that AI exists to augment humans, not replace them and, just as important, that humans are the experts, not the machines. Both notions were expressed continually during the conference. But the 911 community’s desire to learn more clearly has exploded.

Numerous use cases already have emerged and were discussed during the workshop and other educational sessions, and many more are sure to follow as AI technology evolves and the public-safety sector learns how to use it. They include:

• Data analytics — Algorithms are used to parse and contextualize tremendous amounts of data in real time Examples of how this could benefit 911 centers include enabling officials to rapidly determine why call-handling performance has diminished and supervisors to quickly relieve telecommunicators after they fielded multiple high-stress calls in a short timeframe.
• Transcription and translation — The ability to transcribe 911 calls and radio transmissions seamlessly in real time delivers several benefits. Transcriptions can flag keywords and voice characteristics, such as inflection and tone, to help supervisors get in front of potential telecommunicator mental-health issues. They can be reviewed by officials to improve quality assurance and training programs. Meanwhile, translation helps telecommunicators more effectively handle calls placed by non-English-speaking people. Language translation is a big deal — in the current environment, achieving it can take several minutes, which is an eternity during an emergency when every second matters — in fact, in one AI-related sessions during the conference, it was stated that reducing response time by one minute can save 10,000 lives annually.
• Faster content generation — Generative AI can turn myriad telecommunicator and supervisor notes into cohesive reports — much faster than humans can generate them — that can be used to improve 911 center operations. The solution also can generate scenario-focused scripts that can be used to enhance training regimens and to write protocols designed to streamline and enhance operations.
• Nonemergency call handling — It is estimated that as much as 80 percent of the daily call volume handled by 911 centers are nonemergencies, e.g., calls to report incidents like trees felled by a major storm and traffic lights that aren’t working, or to get information about local events. Many of those calls are placed on 10-digit administrative lines and some agencies are using AI to handle them, enabling telecommunicators to focus on actual emergencies, which reduces their stress, improves call-handling times, and results in faster, better-informed dispatching of emergency response.

Jeffcom 911 is embracing the use of AI to handle nonemergency calls. The consolidated 911 center serves Jeferson County, Colorado, which is located along the Front Range of the Rocky Mountains and adjacent to the city and county of Denver; its county seat is Golden, which is home to the Coors Brewing Company. Jeffcom 911 fields about 750,000 calls annually — 250,000 emergency and 500,000 nonemergency — and dispatches for 30 law-enforcement, fire/rescue and emergency medical agencies.

Michael Brewer, the center’s deputy director, shared during the workshop how AI is reducing pressures stemming from a staffing shortage, a common malady across the 911 community. “I don’t have enough people to handle those calls,” he said.

Brewer described the AI solution that Jeffcom911 is using as a force multiplier that has helped telecommunicators focus on emergency calls and improved their performance. He added that, prior to launching AI-driven chatbot in December 2022, Jeffcom 911 had gone 22 consecutive months without meeting NENA’s standard that calls for 90 percent of 911 calls to be answered within 15 seconds. In 2023, the center met the standard in six months and came very close to meeting it in two other months.

According to Brewer, the customer service provided to actual emergency calls “has skyrocketed,” and the center has received no complaints from callers handled by the chatbot. “If it takes someone a few more seconds to get connected to animal control, I’m okay with that,” he said.

Moreover, the return on investment has been spectacular; the cost to procure and implement the solution — which handled about 300,000 nonemergency calls last year — was roughly equivalent to the cost of hiring one full-time equivalent. “The cost is negligible,” Brewer said. “I can’t think of a good reason to not do this.”

MCP’s Kelly described the use of AI in the public-safety sector, especially in the 911 community as “transformative” because it enables agencies to do more with less — in part by relieving telecommunicators from repetitive and mundane tasks, in addition to handling nonemergency calls — and offers unprecedented knowledge augmentation. But he cautioned agencies to avoid the “shiny object” phenomenon.

“You want to be careful with AI because you want to make sure that you’re using it to solve an actual problem,” Kelly said, adding that such solutions shouldn’t be implemented unless they can be operationalized, which means first understanding the need, ensuring that the solution aligns with the need, and then developing policies and procedures that govern the solution’s use.

John Persano is senior business-development manager for 911 for Amazon Web Services (AWS), which developed the Amazon Connect chatbot. Currently, 17 early-adopter ECCs have deployed Amazon Connect and are using it to handle nonemergency calls. The results have been astounding, with a 20 percent to 50 percent reduction in nonemergency administrative calls being answered by telecommunicators. Nevertheless, Persano also believes that public-safety agencies should proceed cautiously concerning AI implementation.

“Our recommendation is to take it slow, understand the benefits, understand the risks, and understand the technology,” he said. “Just because you can doesn’t mean you should.”

After an agency has implemented an AI solution, it must teach its personnel not to blindly trust the outputs that are generated, suggested Dr. Keri Stephens, a professor at the University of Texas – Austin, who is researching AI and its potential applications to support emergency-response operations. This happens largely because of self-doubts, which are common amongst humans.

“People start to think that the machine is right, and you have to guard against that,” Stephens said. “Humans are the experts, not the machine.”

Data bias is a big reason why AI outputs shouldn’t be blindly trusted. It occurs when datasets contain erroneous or incomplete data.

“When we’re talking about training a machine to do something, the quality of the data definitely matters, but so does the quantity,” said Doug Gartner, an AWS solutions architect. “If you have data that’s really high quality but you only have five data points, there’s not enough variety and disparity to see what the machine wants to be able to see. … To emphasize this point, the old paradigm of ‘garbage in, garbage out,’ still holds true.”

Apple announces Emergency SOS enhancements — Trey Forgety, the former government affairs director for NENA and now a software engineering manager at Apple, announced that the company will add new capabilities to iPhone, version 18 and Apple Watch, version 11 that are scheduled to launch in the fall.

When these devices detect that a crash has occurred and the victim is able to communicate but is unable to interact with the device — perhaps because they are injured or they have been separated from it — 911 telecommunicators will be able to pause the call recording and receive an updated location — including an estimated search radius — simply by pressing their call-handling system’s DTMF (dual-tone, multifrequency) button. The location capability is especially advantageous for crashes where the vehicle has left the roadway and is difficult to see due to terrain and foliage.

In addition, a new iPhone feature makes it quick and easy for telecommunicators to request alive video streams from a device’s camera. “There’s no app to switch to, no link to click, and no settings to change,” Forgety said. The telecommunicator sends a sharing request to the iPhone; if the user accepts the request, by clicking on the new “sharing” button, the video stream will be transmitted. An example presented concerned a telecommunicator providing cardiopulmonary-resuscitation (CPR) instructions; the live stream enabled the telecommunicator to coach regarding hand position and whether the compressions were being administered properly.

Telecommunicator reclassification needs grassroots push — Brian Tegtmeyer, coordinator of the National 911 Program within the National Highway Traffic Safety Administration (NHTSA)’s Office of Emergency Medical Services, addressed the ongoing effort to get the Bureau of Labor Statistics (BLS) within the Office of Management and Budget (OMB) to reclassify 911 telecommunicators, who are classified as clerical workers and not as first responders, which would be far more appropriate given the skillsets required for the job. BLS reviews the Standard Occupation Classification system about every 10 years, with the next review scheduled for 2028.

Two bills that would resolve this have been introduced into Congress: the 911 SAVES Act (in 2021) and the Enhancing First Response Act (late last year). However, while both have been referred to committee, neither has moved forward, and it is unlikely that they will in the short term given that this is an election year and Congress has more pressing priorities right now, like passing a budget.

Some believe then that the upcoming lame-duck session, which begins after the election and concludes on January 20 when the next Congress is seated, might offer some hope, because that’s when noncontroversial bills tend to get acted on — and the reclassification bills certainly qualify and in fact seem to have strong support (e.g., the 911 SAVES act has 133 cosponsors). However, others believe that Congress might be reluctant to act because it won’t want to circumvent the OMB/BLS process. Tegtmeyer is one of them.

“One of the reasons that Congress has not passed the 911 SAVES Act, in all of its previous iterations, is that some members hold to a truth that we have rules, and we have procedures, and we have policies, and that what the BLS/SOC already is,” he said.

Tegtmeyer added that despite the fact the SOC review is a slow process that ensues over a 10-year cycle, many members of Congress don’t want to sidestep BLS and legislate the reclassification.

“I think that’s the only reason that something so popular hasn’t passed before,” he said.

Consequently, Tegtmeyer believes that the better approach for the 911 community is to not hope for Congress to act but instead develop a new advocacy strategy that is different from those of the past. The first step is to recognize that BLS is not the typical federal agency.

“BLS is just a data-driven organization; it is a statistics-driven entity, … When we looked at this 10 years ago, there was a groundswell of movements trying to talk about the nature of the job, and the stress and it was emotion-filled,” he said. “But they don’t make decisions based on emotion or anecdotes — they make decisions for data, with data.”

For that reason, Tegtmeyer advised every 911 center to update its job descriptions as soon as possible. BLS will be surveying 911 centers as it reviews the telecommunicator classification , he said, and while it won’t be reaching out to every center, the more centers that have job descriptions that truly reflect the rigors of the job and the skillsets needed to perform it, the better.

“Are your telecommunicators using a typewriter or are they sitting in front of six, eight, or 10 monitors where they’re utilizing computer-aided dispatch? Are they using digital call-handling equipment, accessing CJIS databases, or activating tornado sirens? All these different things that are in your workflows need to be documented in a variety of places.”

In 2022, the National 911 Program created the Developing a Public Safety Telecommunicator Job Description tool kit, which can be accessed by clicking here.

Beyond the reclassification effort, accurate job descriptions have another important, intrinsic benefit, according to Tegtmeyer.

“If you don’t have accurate descriptions, you do a disservice to both the agency and the applicant,” he said. “If you have an antiquated job description … it’s not fair to the applicant because they won’t understand what they’re applying for. Job descriptions need to reflect the realities of the job.” He further suggested that 911 centers review descriptions every time they post a job to ensure that they’re still accurate.

Glenn Bischoff is MCP’s content specialist. Email him at GlennBischoff@MissionCriticalPartners.com.