Artificial Intelligence: What it Means for the Fire Service

Artificial Intelligence: What it Means for the Fire Service

I hope this article finds you well!

Artificial intelligence (AI) is here, and its impact is felt across nearly every part of our lives. You may even recognize the growing presence of AI in the familiar phrase “I hope this email finds you well!”, a staple opening for AI-driven emails and messages, especially when language models first appeared. Initially, AI was touted as a powerful tool for generating reports and handling correspondence, and at the very least, it has certainly help with catching simple grammar errors and mispellings.

As AI has grown more advanced, public opinion about its potential has swung to extremes. Some people view AI as overrated, questioning its authenticity and doubting its practical value. On the other hand, some see AI as a potential harbinger of a dystopian future, a force that could end up controlling human lives in unpredictable and possibly threatening ways. The reality likely falls somewhere in between these two perspectives. AI is still developing, and while it does face limitations that may slow its progress, it has undeniable strengths that could prove valuable to the fire service if harnessed properly.

What is AI?

Here’s a simple explanation of AI with the caveat that there are people far more qualified to describe it. I’ll do my best to explain it in terms that make sense to me. At its core, AI functions as an incredibly powerful prediction machine. Similar to the text prediction on your phone, AI uses patterns to determine what might come next. AI does this by looking at massive amounts of data, sometimes as extensive as the entire internet. Companies like OpenAI have invested billions in creating these enormous models that can learn from patterns and use probability to predict the next image, text, or data point.

AI operates using techniques like Machine Learning (ML) and Deep Learning (DL), which allow it to analyze data and make predictions:

  • Machine Learning (ML): In machine learning, algorithms are trained on historical data, enabling them to recognize patterns and predict future outcomes based on what they’ve seen before.
  • Deep Learning (DL): Deep learning uses neural networks with multiple layers, allowing AI to process complex patterns. This subset of ML drives the more advanced applications, like image and voice recognition, by simulating the neural connections of the human brain.

This high-level pattern recognition allows AI to simulate reasoning. It appears to “think” because it draws from far more training data than any human could process, predicting outcomes with high accuracy based on what it has learned. But here’s the catch: what we call “reasoning” is often a much simpler task for humans than it is for AI.

Moravec’s Paradox describes this irony well: the tasks we humans find the simplest, like physical movement or intuition, are often the hardest for AI, while complex cognitive tasks like math or strategic planning come more naturally to it. (You can read more on this paradox here: Moravec’s Paradox .) While AI may become proficient at sifting through large volumes of information and observing trends, it will likely struggle when the fire service needs quick, real-world decisions and decisive action.

Limitations of AI

In the book Sources of Power, Gary Klein describes a compelling scenario that illustrates one of AI’s fundamental limitations. A battalion chief faces a situation where a vehicle is teetering dangerously on the edge of a bridge. He has never encountered this exact scenario, yet he begins solving it by mentally simulating various solutions in his “mind’s eye.” Drawing from past experiences, training, or even ideas that seem to appear from thin air, he tests different actions in his imagination. Each solution gets mentally rehearsed and revised until he feels one approach might work best. Once he identifies the most promising path, he tries the first step in real life and watches how it plays out, all while mentally preparing for the next stage.

AI, however, lacks this ability to “imagine” specific actions in real-time. It doesn’t have a mind’s eye that can visualize hypothetical scenarios or test creative solutions based on intuition. Instead, AI might generate a broad list of possible steps based on its training data. However, it won’t be able to predict the exact sequence of specific actions necessary to adapt to the evolving conditions of the scenario.

In fact, if you ask an AI like ChatGPT to advise on the "car hanging off the bridge" scenario, you might get something along the lines of, “In a vehicle rescue scenario, stabilize the vehicle, assess occupant safety, and remove the individual using proper equipment.” While this answer covers the general principles, it lacks the real-world, adaptive imagination needed to visualize a unique rescue solution for a vehicle hanging precariously from a bridge.

Beyond a lack of imagination, AI is unable to grasp the intricate realities of human emotion, bravery, and duty. Decisions in emergency scenarios are often shaped by a blend of courage, empathy, and an awareness of the high stakes involved. Firefighters operate with their own set of values, honed by experience and a sense of duty to save others. An AI, however, would consistently default to the safest and most predictable solution based on probabilities and training. It wouldn’t consider taking a higher risk, such as lowering a firefighter for a quick, daring rescue, because its programming is inclined toward minimizing risk rather than embracing it for the sake of saving lives.

Human decision-making often involves factors AI cannot process: a gut feeling, deeply ingrained values, and the flexibility to change course mid-action based on what “feels right.” These uniquely human qualities can lead to extraordinary decisions that AI would find incomprehensible.

Some advancements might make it seem as though AI is approaching human-level decision-making, as demonstrated by recent news of AI-piloted planes outmaneuvering human pilots in dogfights. (For more, see this article on DARPA’s AI-controlled plane .) But in these cases, AI operates within strict parameters, making decisions based on airspeed, altitude, and other data to optimize each move. It’s accessing large amounts of training data to decide on the best maneuver within a tightly defined set of conditions. However, if an out-of-the-box decision was needed, like one driven by empathy or moral complexity, the AI would be at a loss.

Consider the story of German pilot Franz Stigler and American pilot Charlie Brown during WWII. Stigler, upon encountering Brown’s severely damaged B-17 bomber, could have easily shot it down. Instead, he chose to escort the crippled plane to safety, guided by a sense of empathy and humanity that overrode protocol. AI lacks the emotional intelligence to make such a decision; it would be bound to its programming, with no room to consider empathy or ethical nuance. Additionally, whether this decision was “right” or “wrong” depends on perspective, something AI isn’t equipped to grasp.

Creativity is another area where AI falls short, largely because AI “knows” too much. Humans often find creative solutions in ambiguous situations precisely because they don’t have all the answers, allowing them to think outside the box. In my work on a vaccine hub, one of the reasons we were able to find innovative solutions was because we’d never built a vaccine hub before. It was the absence of a clear blueprint that drove us to experiment and find creative ways to solve logistical challenges (more on this in my LinkedIn article ).

When too much data is available, AI often defaults to known solutions, missing out on creative, adaptive ideas that ambiguity can inspire. In situations that require imagination, flexibility, and an ability to step beyond rigid protocols, AI is limited by its programming, and this is where human intuition will always have an advantage.

What Can AI Do for the Fire Service Despite Its Limitations?

Given AI’s limitations, its inability to imagine novel solutions, understand human emotions, or create spontaneously, what value does it bring to the fire service? Despite these constraints, AI has unique strengths that can assist with certain tasks, especially when applied strategically.

One area where AI’s capabilities align well with the fire service is in the OODA loop —Observe, Orient, Decide, and Act—a decision-making process originally developed by military strategist John Boyd (no relation). AI excels at supporting the first two steps of this loop: Observe and Orient.

  1. Observe AI is incredibly efficient at gathering and processing vast amounts of data in real-time. In emergency situations, this ability could be a game-changer. For instance, AI can monitor multiple communication channels, weather reports, and historical incident data, synthesizing it into a coherent picture much faster than human teams alone could manage. By acting as an “observer” on our behalf, AI can help us see more of the situation as it unfolds.
  2. Orient Once the data is gathered, AI can organize and prioritize it, helping incident commanders quickly understand the most critical aspects. During high-stress incidents, such as an active shooter event, important information can easily get lost in the noise. In our own experience, early in an incident, an officer radioed that the shooter was down, but that critical information was lost in the chaos of the moment. An AI could have filtered through the communications, flagged this as “primary” information, and brought it to the commander’s attention. Knowing that most active shooter incidents involve only one perpetrator, AI could assist in making swift, risk-based decisions based on probabilities, which would help focus resources and reduce unnecessary risk to responders.

The faster we can move through Observe and Orient, the more time we have to take meaningful action at the beginning of an incident when each second counts. Quick initial actions are essential because as incidents evolve, our ability to influence them often diminishes. Later interventions tend to be less effective since they’re responding to a more developed and possibly more dangerous situation. In my article on time in emergencies, I discussed how “hunting for seconds” early on can make all the difference in an emergency (for more, see Time Keeps Ticking ).

While AI could have brought clarity and insight to our active shooter scenario, it’s important to note what it couldn’t do. In this situation, several courageous individuals chose to treat patients in the “hot” zone, fully aware of the risks. This decision, driven by human bravery and a sense of duty, is something AI could neither recommend nor understand, as it can’t perceive risk in the same way a person does.

Key Capabilities to Look for in AI for the Fire Service

As we look toward future applications of AI, here are some areas where AI’s strengths could be particularly beneficial to the fire service:

  1. Real-Time Data Analysis: AI can rapidly process environmental, incident, and resource data, giving teams a faster, more comprehensive view of an evolving emergency.
  2. Incident Command Support: AI can assist in tracking units, monitoring communications, and flagging critical information, helping incident commanders stay focused on primary issues.
  3. Predictive Monitoring: AI could predict risks by analyzing data trends—like weather changes, traffic patterns, and resource availability—allowing for proactive rather than reactive responses.
  4. Resource Allocation Optimization: AI could analyze historical and live data to recommend optimal placement of resources based on patterns, helping teams make quicker, more effective decisions.
  5. Administrative Support: AI bots could streamline access to policies, forms, and procedural information, allowing personnel to retrieve information on demand, saving time for critical tasks.

These are just some of the areas where AI’s unique strengths could complement human expertise, providing support that allows our teams to act faster, more decisively, and with better situational awareness.

Current AI Projects in Development for the Allen Fire Department

Our organization is actively exploring AI technologies to support our personnel and enhance situational awareness. Here’s a look at some of the AI projects currently under development:

  1. Reference Assistance Guide (RAG) with Microsoft Azure One of our key projects is developing a Reference Assistance Guide (RAG) using Microsoft Azure. This AI-driven bot will give personnel easy access to policies, procedures, and infrequently used information. For instance, a firefighter could ask the RAG bot a question like, “How do I use FMLA?” and instantly receive the relevant response, along with links to policies and forms. By providing on-demand answers, the RAG bot will allow personnel to focus more on critical tasks, reducing time spent searching for information.
  2. Watchguard AI for Micro-Level Weather Monitoring Another exciting project is our “Watchguard” AI, which continuously monitors weather feeds and prediction models. By linking weather data with Geographic Information System (GIS) data, Watchguard AI assesses potential dangers at the micro level, helping us understand how changing weather conditions could impact the community. Whether it’s anticipating severe weather, road flooding, or high-wind conditions, this AI will give us the capability to proactively prepare for and respond to emerging risks.

Other Promising AI Technologies for the Fire Service

There are many products on the horizon that show significant promise for enhancing fire service operations. Here are a few examples of innovative AI applications already in development:

  • 911 Call Analysis AI is being designed to listen to 911 calls, analyzing voice tone, urgency, and background noise to assess the severity of the situation. These insights could help dispatchers determine the appropriate type and capacity of response units, leading to faster, more tailored deployments.
  • Incident Command Assistant Tools like 3am Innovations’ FLORIAN are being developed to act as a real-time assistant for incident commanders. FLORIAN tracks units, listens to radio communications, and highlights critical information. This technology can prevent important details from slipping through the cracks and help the Incident Commander (IC) make more informed decisions. (Learn more about FLORIAN at 3am Innovations .)

Infrastructure Needs to Support AI

As these AI tools become more integrated into emergency response, we’ll need robust infrastructure to support them effectively. Here are two key areas where future improvements will be essential:

  • Cellular Capacity To leverage AI’s full potential, our data infrastructure must evolve. Moving large amounts of data quickly will soon be as crucial as our radio systems. High-speed, high-capacity cellular networks like 5G are essential for supporting AI in real-time applications. T-Mobile For Business T-Priority initiative is a promising step in this direction, aimed at meeting the growing data needs of public safety. (more about T-Priority here .)
  • Edge Computing Edge computing allows data to be processed close to its source rather than relying on distant data centers. For the fire service, this would mean faster access to critical insights at the incident site, even if traditional networks are down or limited. Edge computing will enable real-time processing, providing vital information on-site without dependency on external servers.

Final Thoughts

Being on the frontline of AI’s evolution is both challenging and exciting. When the internet first emerged, no one was sure of its future applications, but over time, it transformed how we communicate and work. I believe AI will follow a similar trajectory, gradually integrating into our workflows and ultimately enhancing our effectiveness. As we continue to implement AI in the fire service, we’ll gain tools to make our communities safer and our teams more efficient.

Further Reading

For those interested in exploring AI’s broader impact, here are some recommended resources:

  • Altman, S. (2023). “The Age of Giant AI Models Is Already Over.” Wired .
  • Klein, G. (1998). Sources of Power: How People Make Decisions.
  • Brynjolfsson, E., & McAfee, A. (2017). The Business of Artificial Intelligence. Harvard Business Review.
  • Agrawal, A., Gans, J., & Goldfarb, A. (2018). Prediction Machines: The Simple Economics of Artificial Intelligence. Harvard Business Review Press.
  • Davenport, T. H., & Kirby, J. (2016). Only Humans Need Apply: Winners and Losers in the Age of Smart Machines. Harper Business.
  • Wilson, H. J., & Daugherty, P. R. (2018). Collaborative Intelligence: Humans and AI Are Joining Forces. Harvard Business Review.
  • Ransbotham, S., Kiron, D., Gerbert, P., & Reeves, M. (2017). Reshaping Business with Artificial Intelligence. MIT Sloan Management Review.

Greg Schon, Architect, LEED AP

"WE SHAPE OUR BUILDINGS, THEREAFTER THEY SHAPE US" -Winston Churchill - Senior Designer of City and County Buildings including City Halls, Fire and Police Stations, Courthouses and other Municipal Structures.

1 周

What a great analysis of the strengths of AI input as well as weaknesses in how current AI processes data and potential scenarios! Thanks for this!

Travis Green, Ph.D.

Assistant Fire Chief | Lecturer | Agile Fanatic and Tech Enthusiast!

2 周

This is a great article on how AI can be used in the fire service. I have been very interested in this and am glad to see the Allen FD being innovative in its use! As part of your real-time data analysis, live video analytics using AI on-scene will be fascinating in its analysis. As you stated before, this will all be taxing on current data networks.

Pavel Uncuta

??Founder of AIBoost Marketing, Digital Marketing Strategist | Elevating Brands with Data-Driven SEO and Engaging Content??

2 周

Interesting perspective! AI can enhance data organization for quicker response. Human touch still vital. Exciting to see tech evolve! ?? #AI #EmergencyResponse #Innovation

要查看或添加评论,请登录

社区洞察

其他会员也浏览了