By Jim Brown
No doubt you’ve heard about artificial intelligence and how it’s transforming the world. But what does that mean for the fire service and how will it affect you? Let’s start at the beginning.
What is AI? A brief history lesson
AI is a general term to describe intelligence derived by machines instead of humans. The study of machine intelligence is actually not new. The concept of machine intelligence was first introduced by mathematician Alan Turing in 1950, building on work from a 1943 paper on artificial neural networks by neurophysiologist Warren McCulloch and logician Walter Pitts. Dartmouth researcher John McCarthy is credited with the first use of the term artificial intelligence in 1956.
Research into machine learning and neural networks continued with varied success through the 1960s and 1970s. In the 1980s and 1990s, advances made machine learning a practical method for training robotics. At the same time, work on neural networks and transformers progressed slowly, with initiatives being started and stopped based on their limited success.
By 2018, Google and OpenAI had transformer-based systems functioning with mixed results. That all changed in November 2020 with the release of OpenAI’s ChatGPT, the first commercially successful transformer-based chatbot – a software application or web interface that allows for textual conversations between a person and computer.
AI at the fire department
Today the term artificial intelligence is used to describe a broad field of study and application in robotics and intelligence. It is generally divided into narrow AI and artificial general intelligence (AGI):
- A narrow AI is a system that performs a single task, like AlphaGo mastering the board game Go. According to a 2024 Stanford University study, seven of the nine narrow AI systems reviewed exceeded human capabilities in their field. So yes, they’re smart.
- A true AGI system would approximate average human intelligence across all fields of knowledge. Some researchers believe we will achieve AGI in 2025. I believe we are already there.
What’s getting all the attention lately is a type of general-purpose narrow AI called generative pretrained transformers (GPT) – that’s the GPT in ChatGPT. The most well-known chatbots include Claude, ChatCPT, Gemini, Copilot and Perplexity. These systems have been trained on the breadth of human knowledge by scraping the internet (searching and absorbing its information) and then being fine-tuned by humans to so their responses align with typical human responses.
These systems, also known as models, excel at language and pattern recognition. Those skills are incredibly valuable to the fire service.
AI can streamline operations in a variety of areas for fire and emergency services. Some examples: incident briefings for the PIO, CRR outreach materials, multi-language public education materials, staff reports, document analysis, evaluating policies for consistency and aggregating survey results, to name just a few.
How to implement AI at the fire department
There are several factors to consider as you ponder how to implement AI into your workflow. I have simplified them into five categories:
- Change management: Change management is necessary any time we introduce significant operational or administrative change. We must communicate our intentions, involve stakeholders, and provide training and support. I recommend appointing an AI strategist as your change facilitator.
- Data preparation and security: Data preparation is important for AI to effectively access our data in a safe and efficient manner. This includes thinking about file-naming conventions and folder structures, as well as internal- and external-facing data. Data preparation helps with data security, which means keeping private data private!
- Use cases: Use cases can be simple or complex. We are seeing some complex AI integrations in wildfire detection, for one. (See the list above for some examples.)
- De-risking threats: De-risking threats involves identifying area where AI can cause harm. There’s a lot to consider here. Some considerations include providing transparency and monitoring for bias and errors.
- Governance: Governance is the term I use to describe the policies and plans we adopt to encourage ethical and transparent use of AI.
Note: This is not a sequential list where you must proceed from step 1 to 5; the elements need to be considered concurrently.
An AI case study
To pull it all together, let’s work through a scenario, which typically starts with a use case: You have played around with one of the chatbots at home, considering the various ways it could help you at work. You decide you want to have it help you write a policy. Great use case! Follow these questions to work through the other four factors:
- What is your normal policy development process (change management)? How do we communicate that this is a test case vs. a permanent change? How do you make sure all stakeholders are involved? What are your stakeholders comfortable with AI doing? Make sure you are not sending the message that it’s open season for AI!
- How do you keep your data safe (data preparation and security)? What information can we give the AI to work with? Are we dealing with any sensitive data? Make sure your IT department is one of your stakeholders, as they will address the data preparation and security pieces.
- What threats does this opportunity present (de-risk threats)? How do we protect ourselves from the unknown unknowns? Have we provided any training on these AI tools? Remember, educate to innovate!
- How do I protect our organization, and what do I need to disclose about how I used AI (governance)? I cannot emphasis enough the importance of having an AI use policy. A good AI policy will address these factors and many more considerations for the effective, transparent and ethical use of AI. A good policy will guide you through implementation.
What sounds like a simple decision to use AI to draft a policy is actually more complex. But you can solve a lot of these challenges with a little upfront work. Get involved with local and national AI groups. The GovAI Coalition is a great resource for policy templates and training resources. And remember, don’t be scared of AI – be deliberate!
Final thoughts
AI is advancing faster than any technology in human history. It’s only getting smarter, but it still makes mistakes. You need to maintain the “human in the loop” approach. Verify outputs. Have backup processes. Think of AI as your creative partner. It relies on you to maximize its capabilities. And if you find something that AI can’t do, the correct mindset is likely, AI can’t do that yet.
This is my original work, reviewed by Claude chatbot and supported by one perplexity search to get the timeline for AI.
WATCH | ‘Do the next right thing right’: Kevin Sofen talks tech fundamentals
ABOUT THE AUTHOR
Jim Brown is a retired division chief from Monterey, California, now living in Hawaii. He is a California-certified Master Instructor, a member of the IAFC Technology Council’s AI subcommittee, and a contract instructor for “Analytical Tools for Decision-Making” at the National Fire Academy.
