Dan Chuparkoff challenges fire service leaders to use AI to reduce routine work while keeping judgment, accountability and decision-making human
AI should be viewed as an assistant, not an automator. That was the key message from technology executive and AI educator Dan Chuparkoff speaking to fire service leaders at Fire-Rescue International 2026 in Kansas City, Missouri.
During his presentation, “AI & the Future of Fire & Emergency Services,” Chuparkoff explored how generative AI works, why it will continue to make mistakes and where it could change fire service operations. He also cautioned against handing over decision-making that requires experience, context, ethics and accountability. The ultimate goal, he said, is to learn how to use AI for portions of our work that already consume too much of their time, freeing us to focus on bigger picture strategies and tasks.
The next copy-and-paste moment
Chuparkoff grounded the presentation in a story from his first job as a 17-year-old working in the architecture and engineering division of a Florida school district. His assignment was to draw parking spaces by hand. Each line had to be measured and drawn individually in ink on Mylar. Completing a parking lot could take several days.
Then he met the only architect in the building using AutoCAD. Chuparkoff watched him draw one parking space to the correct dimensions and angle, then copy and paste it repeatedly. Work that had taken Chuparkoff days was completed in about 90 seconds.
“You can’t go back to the old way after seeing productivity help that’s that great,” he said. “It would feel absurd.”
His boss resisted the technology, fearing the computer would take architects’ jobs. But that fear was based on the assumption that AutoCAD would independently design buildings, solve complex architectural problems and make decisions without an architect — something it would never do. Rather, it simply made a repetitive part of the architect’s work faster.
Simply put, his boss saw AutoCAD as an automator. Chuparkoff saw it as an assistant.
Chuparkoff described AI as the latest in a series of workplace shifts that included personal computers, spreadsheets, the internet, mobile devices, cloud computing, data science and remote work. Each technology eventually became so integrated into daily work that people stopped talking about it as something separate. He believes AI will follow the same path.
AI predicts; it does not know
To explain generative AI, Chuparkoff asked the audience to complete the phrase “Once upon a time.” The response was immediate and nearly unanimous.
That exercise demonstrated what generative AI does: It predicts the most likely next word based on patterns in the information on which it was trained. When generating an image, it performs a similar calculation to predict the next pixel.
The method makes AI fast and capable, but it also creates limitations. AI tends to produce the consensus answer rather than the unusual, highly specialized or genuinely original one.
Chuparkoff characterized AI as a “B-minus student” across a wide range of subjects. It can raise a user’s baseline ability in areas outside that person’s specialty, helping a fire officer become a better researcher, planner or communicator. It does not make the system an expert in emergency operations.
Fire service professionals should remain the “A-plus experts” in their field, he said, applying their training, experience, judgment and ethical standards to everything AI produces.
The distinction becomes more important as an incident or question becomes less routine. AI may perform well when asked about a topic covered repeatedly online. It is less dependable when asked to evaluate an unusual emergency with limited information and serious consequences.
Underneath every answer is a probability calculation. The system may select one answer because it has a 40% probability of being correct, even though 40% confidence would be unacceptable for the decision facing an incident commander.
“The weirder the thing is that you’re asking for help with, the worse AI is going to be at solving that problem,” Chuparkoff said.
Wrong answers are part of the system
Hallucinations are often treated as a temporary defect that technology companies will eventually eliminate. Chuparkoff argued that the problem cannot be separated entirely from how generative AI works. After all, he explained, “The internet is filled with sarcasm. It’s filled with trolls deliberately trying to mess stuff up. It’s filled with jokes. It’s filled with experts, and it’s filled with novices. The experts are dramatically outnumbered.”
Chuparkoff pointed to a now-infamous online suggestion from a Reddit user to add glue to pizza sauce to keep cheese from sliding off. Because users had elevated the answer online, an AI system later repeated it as legitimate cooking advice. The system accurately reflected information it found. The information itself was bad.
AI output requires review. The level of review should increase with the stakes, the novelty of the situation and the consequences of being wrong.
Chuparkoff compared that relationship to autocomplete. People have used autocomplete for years, accepting suggestions that fit and rejecting those that do not. Few would allow autocomplete to send messages to friends, employees or residents without first reviewing them. The same principle should apply to more advanced AI systems.
Humans maintain what Chuparkoff called a “context advantage.” They know what happened at breakfast, what the chief said yesterday, what failed during the last incident and what the organization is trying to accomplish next year. AI does not know any of that unless someone provides the information.
Human decisions are also shaped by memories and hopes — lessons from past experiences and ideas about a future that does not yet exist. AI can contribute information to those decisions, but it should not own them.
Where AI could change fire service work
Chuparkoff encouraged attendees to begin by asking an AI assistant how the technology could change their specific roles, services and constraints. For his demonstration, he asked for nine ways AI could affect the work of fire chiefs, commissioners, department leaders and firefighters responsible for emergency response and community risk reduction while facing staffing shortages, tight budgets and fragmented data.
The point was not the nine-point list. It was to encourage fire service leaders to start asking these types of questions now, rather than wait for every records system, training platform or response tool to release a built-in AI feature.
“I don’t care if you agree that those are the nine things or not,” he said. “My message is that you should start a conversation with your AI assistant.”
That conversation still requires scrutiny. Departments will need to address data quality, privacy, security, accuracy and who remains accountable for the final action.
Use AI as a coach
One of Chuparkoff’s most practical recommendations involved changing the sequence in which people use AI. Specifically, he described asking AI to write a LinkedIn post for him. The result was generic, repetitive and included information that was not true. Repeated prompts to make the post sound more like him did not solve the problem.
So he reversed the process. Chuparkoff wrote the post himself, then gave AI the new draft and examples of his previous work. He asked the system to rank the new post against the others and explain why it did or did not belong among his best.
He now applies that process to customer calls, project plans, research briefings and presentations. Instead of asking AI to replace his first effort, he asks it to assess what he did and identify where he could improve.
“Use AI to make you better,” he said. “AI can coach. AI can assess what you just did and give you feedback.”
For fire service leaders, that could mean asking AI to review a training plan against previous plans, identify gaps in a briefing, compare versions of a policy or evaluate whether a written message addresses the questions members are likely to raise. The person still produces and owns the work. AI provides another review.
Flip the hierarchy of work
When addressing how to organize your work — and where to know when to tap into AI, Chuparkoff showed a six-level hierarchy of work.
The three broadest layers at the bottom are communication, process and investigation. People exchange information, create reports, follow workflows and search for problems that require attention. Above those layers are problem-solving, decision-making and imagination.
The lower layers are necessary, but they can consume the entire week. When that happens, leaders spend less time addressing unresolved problems, making deliberate decisions and imagining better ways to serve their communities.
AI is particularly suited to handling information. It can summarize a large volume of material or expand a small amount of information into a more detailed draft. Used carefully, it can reduce time spent reading documents, preparing routine reports, searching previous conversations and organizing meeting notes.
Chuparkoff demonstrated that concept with AI-generated notes from his own keynote, a song created from the presentation and real-time language translation. The examples showed how the same information could be captured, reorganized and delivered in different forms.
For fire departments, this could mean helping responders communicate with residents who do not speak English. AI-supported meeting records could also make institutional knowledge searchable and easier to incorporate into future training.
The main goal of reducing routine work is to move human attention higher in the pyramid. Fire service members will still be needed to solve problems that technology has not encountered, make decisions shaped by local conditions and imagine safer ways to operate. Those responsibilities cannot be separated from experience, values or accountability.
Chuparkoff closed by asking leaders to examine the lower levels of their own work and identify tasks an AI assistant could help compress. The time recovered should then be reinvested where human expertise matters most.
“Flip your pyramid of work,” he said. “Bring your AI assistant in to help.”
FireRescue1 is using generative AI to create some content that is edited and fact-checked by our editors.
Artificial Intelligence
AI-guided satellites could give firefighters a new view of spreading wildfires
West Virginia University researchers are testing a system that would allow satellites to detect wildfires and reposition themselves to provide crews with more frequent data on fire behavior
Artificial Intelligence
CAL FIRE turns to AI mapping to sharpen wildfire planning
AI-powered mapping is helping track buildings, defensible space and evacuation routes across 31 million acres to improve wildfire preparedness and response
Artificial Intelligence
Strategic Scan insights: What fire chiefs are saying about AI
CPSE Center for Innovation researchers Alex Henderson and Alex Fisher break down how departments are using AI, where adoption is lagging and what concerns remain
Communications and Interoperability
Connected but not coordinated: The fire chief’s communications challenge
Connectivity has never been better; the challenge now is filtering the noise, establishing priorities and making better decisions under pressure
Artificial Intelligence
‘The Great AI Lie’: The data confirms our fears
Part 2: Research suggests AI is accelerating work, weakening reflection and creating new demands on the people it promised to help
Artificial IntelligenceFire-Rescue InternationalLeadershipTechnology
