Copilot for the Fire Service: The Power of Artificial Intelligence

BY Scott Roseberry and Sai Narain

To remain relevant in the modern era and effectively take advantage of disruptive advancements, it is critical for leaders to embrace change management. Despite the attention-grabbing doomsday predictions, the undeniable reality is that artificial intelligence (AI) will revolutionize every aspect of the fire service. However, it is worth noting that the fire service has historically been one of the last public-sector entities to adopt new technologies. This article examines the significant challenges confronting the fire service and the substantial improvements AI can bring to our operational capabilities and safety measures.

The Language Modernization of the Fire Rescue and EMS Services

What Is AI?

It’s important to understand what AI is and what it isn’t. AI aims to create intelligent machines that can perform tasks that typically require human intelligence, such as visual perception, speech recognition, decision making, and language translation. AI can be classified into several types based on its capabilities and functionalities.

Large Language Models

Designed to understand, generate, and process human language, large language models (LLMs) provide the foundation for generative AI systems. Their ability to understand and model language enables generative AI models to create novel and coherent textual content.

ChatGPT is a specific implementation of an LLM, developed by OpenAI. It is designed to provide conversational capabilities and engage in dialogue with users. ChatGPT leverages the generative pretrained transformer (GPT) architecture, which has been trained on a diverse range of Internet text to acquire a broad understanding of language patterns.

AI can have a significant impact for the fire service, where quick and accurate decisions can save lives and property. This includes the following:

• Predictive modeling of fire risks.
• Optimizing response times.
• Facilitating training through virtual simulations.

Using AI responsibly in this context means prioritizing factors such as accuracy, reliability, and transparency. Decisions made by AI must be reliable and precise, as inaccuracies can potentially lead to dire consequences. Transparency is vital to ensure that AI’s decision-making processes can be understood and trusted by firefighters and the public alike.

We must consider privacy concerns, as AI systems could have access to sensitive data. There must also be a mechanism for human oversight and intervention, because while AI can aid in decision making, the human element remains essential in dealing with the complexities and unpredictability of real-world incidents. Responsible use of AI in the fire service calls for a balance between leveraging its benefits and ensuring safety, trust, and adherence to ethical standards.

Assisting the Fire Service

AI can assist the fire service through many channels, including the following:

• Communications.
• Incident command.
• Violent fire events.
• Wildland urban interface.
• Transportation.
• Deployment models/standards of cover.
• CAD systems.
• Emergency medical services (EMS).
• Unmanned systems.

Here’s a closer look at each channel and how AI can prove helpful.

Communications

Many see fire department communications as one of the greatest challenges, impacting both nonemergency and emergency interactions. Nonemergency communications have mainly consisted of face-to-face interactions and departmental emails. While text messaging and social media have become more prevalent means of communication, email remains the main form of mass communication for nonemergency situations.

Firefighters often receive so many emails that filtering through them becomes difficult. An unintended side effect to this is that the firefighters could overlook communications that require a timely response. Fire chiefs often say firefighters don’t read their emails. In cases where this is true, the result is poor communication up and down the chain of command. The result? An erosion of morale and workforce performance.

Figure 1. Common Types of AI

Emergency communications consist of two primary means: face-to-face and radio interactions. According to the National Institute for Occupational Safety and Health (NIOSH), communications are often cited in line-of-duty deaths (LODDs) and firefighter injuries as contributing factors. This can be from firefighters on the wrong tactical channel, radio “dead spots” that leave a lost or down firefighter unable to communicate, or simply missed communications from scene distractions. For incident commanders (ICs), the amount of radio communications at a multialarm scene can be overwhelming, leading to many such missed communications.

This can be especially critical in the event of a Mayday. Don Abbott’s research with Project Mayday identified 16 terms that often come up before a Mayday. It is easy for the IC to suffer information overload and miss one or several of these key terms transmitted by interior crews, but it is crucial for firefighter safety and survival that the IC recognize these terms and withdraw everyone from the building prior to a catastrophic event.

Incident Command

Today’s fireground incident commander (IC) works under extremely stressful conditions and must make multiple rapid decisions to protect life and property. Dr. Gary Klein’s research on decision making in the 1980s determined that ICs don’t use what we would consider normal means of decision making. His research showed that ICs use a completely different process called recognition prime decision making (RPDM).

For this process, commanders follow the following steps:

1. Perform a size-up of the scene.
2. Look for cues related to situations they have encountered in the past.
3. Mentally apply tactics to mitigate the situation they are looking at based on prior experience or training.
4. Internally process possible outcomes and decide on the best one.
5. Put those tactics into action and repeat the process.

Dr. Klein suggested that for this to work, ICs must possess four key attributes:

1. A strong situational awareness.
2. High levels of knowledge about strategy and tactics.
3. The ability to mentally conduct possible scenarios in their head (much like playing five steps ahead in chess).
4. High levels of confidence in their decision-making abilities.

For RPDM to work effectively, the IC needs robust experience with a multitude of incidents. The issue with this is that according to the U.S. Fire Administration (USFA), fires are on the decline nationwide. This is a double-edged sword, as fewer fires mean fewer losses to our citizens. Conversely, it means fewer experiences.

The current availability of technology has increased the volume of information available to ICs. This can increase their situational awareness in several ways.

• Commanders can see the interior conditions from signals being sent via thermal imagers.
• Wearable technology feeds the command post with the interior crew’s heart rates, core body temperatures, and air bottle status.
• Prefire plans with X, Y, and Z axis tracking show the actual location of firefighters on the fireground in real time.
• Radio communications and face-to-face communications are also feeding in rapidly.

This improves situational awareness. However, it also causes information overload and creates distracting “noise.”

Violent Fire Events

Firefighting is an inherently dangerous job. When firefighters enter buildings for extinguishment or search and rescue, the forces of gravity, fire dynamics, building integrity, and time are all working against the safety of the firefighters and occupants.

Research on modern fire behavior out of the National Institute of Standards and Technology (NIST) and UL has given us a much greater understanding of the physics behind fire dynamics in buildings. We have a better understanding of concepts like flow path, backdraft, and flashover. Smoke explosions, a relatively new phenomenon, are currently being researched by UL FSRI’s Daniel Madrzykowski, Ph.D., P.E. The research indicates that signs and symptoms prior to these events do exist, but they may be too subtle for people to notice, or our situational awareness of the precursors has been compromised. Many firefighter fatalities and injuries could be prevented if we had better warnings of such events.

Wildland Urban Interface (WUI)

Undeveloped land rich in vegetative fuels meets occupied structures and human development in the WUI. The USFA estimates there are 60,000 communities at risk in the WUI and that number is growing by 2% every year. Between 2002 and 2016, more than 3,000 structures were lost to fires in the WUI.

In 2018, California’s most devastating wildfire, the Camp Fire, destroyed 18,804 structures, burned more than 150,000 acres (about half the area of San Antonio, Texas), caused $16 billion in damage, and killed 85 people. The risk of wildfires is staggering and growing. Humans will continue to build in the WUI areas, necessitating better prediction models, evacuation notifications, and fire-adapted communities.

Transportation

The Insurance Institute for Highway Safety predicts there will be 3.5 million autonomous vehicles on U.S. roads by 2025 and 4.5 million by 2030. Torc Robotics, a subsidiary of Daimler Trucks, started testing long-haul driverless trucking in 2022. Electric fire apparatus are already in production. Fire departments in the United States are using them. It’s just a matter of time before manufacturers offer autonomous fire apparatus to the fire service.

Deployment Models/Standards of Cover

The Commission on Fire Accreditation International defines the standards of cover (SOC) as “Standards of Response Coverage as being those adopted, written policies and procedures that determines the distribution, concentration and reliability of fixed and mobile response forces for fire, emergency medical services, hazardous materials and other forces of technical response.”

Essentially, this addresses the need to get the right resources in place to protect our citizens most efficiently within our financial parameters. For many jurisdictions, this can be a complicated task to complete. The SOC considerations include the following factors:

• Population density of the covered area.
• Building density.
• Risk assessments.
• Current service levels.
• First-alarm assignments for residential and commercial occupancies.
• Establishing a distribution criterion.
• Evaluation of in-district reliability.
• Evaluation of historical performance.

This is not an exhaustive list of all the data points needed to produce a well-defined SOC; it’s just meant to show the complexity involved.

Computer-Aided Dispatching (CAD) Systems

Used by public safety answering points (PSAP) to prioritize calls to 911, CAD systems locate available resources and properly dispatch the correct units. CAD systems are the start of the data intake systems fire departments use for KPIs.

Modern CAD systems, Next Generation 911 (NG911), are digital Internet protocol-based systems that incorporate enhanced geographic information systems and the seamless flow of voice, text messages, pictures, and videos from the callers to the PSAP centers.

Even with the advanced NG911 systems, several obstacles still interfere with dispatching the closest available units.

• Operators must be able to interpret the initial caller, who may be in an excited state.
• The caller may speak any number of languages. Translating through a third-party language line adds precious time to the dispatch.
• After identifying the nature of the call, the operator walks the caller through a series of steps to help remedy the situation prior to the fire department’s arrival.
• For medical emergencies, the operator’s identification of the emergency and delivery of correct instructions via phone can greatly impact the outcome.
• During this time, the records management system collects data on the call and the response for later measurements. Much of this information is used for the National Fire Incident Reporting System, soon to be updated to the National Emergency Response Information System.

Unfortunately, complications are also part of the equation. A breakdown in communications and information sharing erodes the safety of the citizens and the responders. And the shortage of trained dispatchers answering 911 calls and accurately dispatching units presents yet another challenge.

EMS

Whether fire departments offer advanced life support (ALS) capabilities with patient transport or provide basic lifesaving capabilities until ALS arrives, firefighters are making critical decisions in short timeframes to improve patient conditions. Firefighters must be able to obtain the following right away:

• A patient history.
• Chief complaint.
• Vital signs.
• A list of current medications and known drug allergies.

From there, they’ll decide on the proper treatment protocol. They must know drug interactions and contraindications as well as understand drug allergies for thousands of brand-name prescriptions as well as their generic counterparts. Staffing shortages around the country mean that firefighters often make these decisions on little to no sleep. The likelihood of a mistake increases with added stress and fatigue.

Unmanned Aircraft Systems (UAS)

No discussion about technology and AI would be complete without acknowledging UAS. These systems benefit the fire service in many areas, including hazardous materials spills, swift water rescues, wildland fire events, and most recently in New York collapsed building rescues.

UAS offer ICs the ability to see what was once unknown until well after the incident was mitigated. The drawback to these systems is that they are staffing intensive and require constant training for effective and efficient use. Although the Federal Aviation Administration has lightened regulations requiring the use of a visual operator (VO), they are still required in certain situations. Few fire departments have resources to devote to flying drones. Staffing shortages have stressed the system while departments still strive to maintain National Fire Protection Association 1710, Standard for the Organization and Deployment of Fire Suppression Operations, Emergency Medical Operations, and Special Operations to the Public by Career Fire Departments (2020).

Using AI Today

Many companies have been incorporating AI technology in their solutions for the past few years. And many of us use AI every day and don’t even realize it. Anytime you ask your phone for directions or the current weather, you are using AI. Here are a few simple, and mostly free, examples of AI solutions that can help you out today:

• Record, transcribe, and summarize a meeting you couldn’t attend.
• Help you put together a standard operating procedure (SOP). (Remember not to share any confidential information with LLMs.)
• Help you enhance an email to look and feel more professional.
• Build out a presentation for which you have the starter information.
• Synthesize data.

Change Management/Fire Service Adoption

The fire service has a strong culture that values experience, established practices, and traditions. Adopting technology can disrupt these and create a fear of the unknown—or a fear of losing long-established workplace norms. It’s becoming imperative for leaders to develop a change-management leadership style to create cultures that are flexible and adaptable to future needs. AI services and software are rapidly advancing, and many technology solutions in use today are already harnessing the power of AI. It will be a disruptive technology just like the Internet and cell phone usage was. However, it will create greater efficiencies and greater capacities to perform lifesaving work. To capitalize on the power of AI and avoid its potential for harm, the fire service will need to make some changes.

Although the mission of the fire department hasn’t changed since its inception, the environment it operates in has changed over the past 30 years. Political ideologies; social responsibilities; reform movements in diversity, equity, and inclusion; and technological advancements have all changed the landscape. Society has also placed greater emphasis on the services offered by its local fire department.

With all these changes, one thing that hasn’t changed for the fire service is the recruitment process. Many of us are still recruiting members the same way we did 20 to 30 years ago. The need for a more diverse work crew and membership versed in the latest technology has increased. The fire service must reconsider its recruitment process to attract this new talent from Generation Z workers.

Leaders need to place greater emphasis on developing a culture that is flexible and adaptable not only for the needs of today but the needs of tomorrow as well. Leaders need to communicate about anticipated changes and disruptions. They will also need to use a change-management process to successfully implement new technologies.

Will AI Replace Firefighters?

Not likely. AI will always have limitations preventing it from replacing humans. As (Chief (Ret.) Alan Brunacini from the Phoenix (AZ) Fire Department said many times, Mrs. Smith is the reason we’re here and why so many of us joined the fire service. AI won’t understand why firefighters stick around to mow the yard of the patient they just transported to the hospital. It won’t understand why driving by a six-year-old child’s birthday party flashing the lights and blowing the sirens sends kids running in all directions screaming in ecstasy. AI will, however, augment firefighters in the future. It will be a companion riding in the jump seat to help ICs make better decisions to improve the effectiveness of their response. It may replace administrative staff for certain tasks, freeing them up to fill in spots in operations and reducing the problems caused by staffing shortages.

Human firefighters will always be willing and needed to answer the call to duty. This article is a starting point for conversations and collaborations on how we will interact with AI. We need to start having those conversations today, as the rate with which AI and other technologies are advancing is faster than we probably believe. Either we start driving the changes today, or they will be driving us tomorrow.

REFERENCES

Abbot, Don. “Project Mayday: Update.” Fire Engineering, 1 June 2020. bit.ly/3KIIJIw.

ChatGPT, GPT-3.5. OpenAI, September 2021. chatgpt.com.

Coffman, Michael. “How AI and Machine Learning are Improving Ambulance Response and Dispatch.” Journal of Emergency Medical Services, 2022. bit.ly/4ecUXq9.

Fagan, Thomas. Center for Public Safety Excellence, 16 April 2015. bit.ly/3yXLfIc.

Gasaway, Richard. “Understanding Fireground Command: Making Decisions Under Stress.” Fire Engineering University, 2012.

Gibbs, Samuel. “How did email grow from messages between academics to a global epidemic?” The Guardian, 7 March 2016. bit.ly/4c7ZE32.

USFA. U.S. Fire Administration, 8 June 2022. bit.ly/4ciqpBU.

Wilde, Elizabeth. “Do Emergency Medical System Response Times Matter for Health Outcomes?” Health Economics, U.S. National Library of Medicine, 2012.

---

Scott Roseberry, EFO, MPA, CFO, is a battalion chief for the Garland (TX) Fire Department with more than 24 years of service. He is the special operations division chief and an adjunct instructor at Tarrant County College. He serves as the vice-chair of the IAFC Technology Council, is a member of the FEMA Region 6 RAC, and contributes to the Department of Homeland Security First Responder Resource Group. He has delivered several presentations on technology and leadership at regional and national conferences. Roseberry is also a cofounder of Technology Summit International.

SAI NARAIN is a director of technology strategy, specializing in public safety and justice. With more than 20 years of experience in the public safety industry and expertise in the fire, EMS, and 911 technology landscape, he currently works with organizations on envisioning the art of the possible, how technology can be a force multiplier and empower organizations with data sharing, collaboration, and artificial intelligence, as well as the power of the cloud.