Doctrine

Frontline Intelligence

AI and technology for fire, EMS, and emergency services.

Fifty Nine Percent Comfortable, Fourteen Percent Covered

Robert Grand · Battalion Chief who still runs calls

The CPSE Center for Innovation just published the first Strategic Scan on AI in the fire service, giving Fire & EMS a baseline. Fifty nine percent of chiefs say they are comfortable with AI. Fourteen percent have a formal AI policy. That gap is an open lane. Agencies that give people a governed path that can keep up with the models will set the standard.

Researchers Alex Henderson of Marist University and Alex Fisher of CPSE surveyed chief fire officers from accredited departments, guided by a working group that included chiefs from Prescott, Arizona; Charleston, South Carolina; Lexington, Kentucky; El Paso, Texas; Grand Rapids, Michigan; and Miami Beach, Florida. The findings landed in FireRescue1 (https://www.firerescue1.com/artificial-intelligence/strategic-scan-insights-what-fire-chiefs-are-saying-about-ai). Forty percent say their jurisdiction gives them access to AI tools. Two thirds already use AI to proofread and revise documents. Seventy percent have no plans to train on AI. Comfort is climbing. Policy and training are not.

I see that same pattern in my own shop. We have an official department AI locked down through the city Microsoft assistant, a Copilot style tool that counts as access on paper. Nobody uses it for real work. It is not great, and it cannot keep up with current online models. Models move fast and the approved tool lags, so the official channel becomes wallpaper. People still use better tools quietly. Everyone is using AI at work and almost nobody says it in meetings. Staff have not been told they cannot, and they have not been given a usable, up to date approved model. That is comfort without coverage.

The Comfort Without Coverage Problem

Comfort without coverage is how agencies drift into quiet use. People reach for a writing assistant because the work is on their desk. Document revision is low stakes, so the tool spreads through admin before anyone names who owns the rules. The Strategic Scan shows administrative use strongest while formal policy sits at fourteen percent. When two thirds already revise documents with AI and only a thin slice has written guidance, the field is ahead of its own doctrine. Access on paper without a model people will actually use is not coverage.

The first question is whether your agency can name, this week, who owns AI policy. Not a committee that meets next quarter. A person or pair able to draft and hold the line. If that name is blank, comfort rises and coverage stays flat. The second question is whether a short policy now, with a governed path that can keep pace, beats a long policy later after quiet use becomes the culture. A page on what may enter an approved tool, what stays out, who reviews output, and how training starts moves you from gray to governed. Give people something usable enough to use AI out loud.

The Training Gap That Will Decide Trust

Seventy percent of chiefs have no plans to train on AI. That number should sit with Fire & EMS leaders longer than the comfort figure. Comfort without training produces uneven practice. One officer drafts with a tool the agency can stand behind. Another pastes something that should never leave the firewall. Training makes that difference a matter of doctrine, not luck.

The third question is whether your agency treats AI literacy like any other operational skill or like a hobby. If it is a hobby, the fourteen percent with policy will stay the exception. If it is a skill, you start small: approved tools, how they stay current, what data may enter them, how to check output, and when a human must own the call. Henderson and Fisher note that data quality, privacy, and accountability remain real concerns. Those concerns do not get solved by waiting or by locking staff into a lagging model. They get solved by naming them in policy and teaching on a path the department owns.

What to Watch Next

Watch whether departments convert this baseline into written standards while administrative AI remains the main use case. Watch whether the seventy percent with no training plans shrink. Watch whether the forty percent with access close the gap to the fourteen percent with policy, and whether that access means a usable tool.

The bottom line is simple. Name who owns AI policy this week. Write a short policy that treats access on paper as worthless if the approved model cannot compete with what people already use. Give staff a governed path that keeps up so they can use AI out loud. Quiet use is the signal. Coverage is the answer.


Robert Grand is a Battalion Chief at Eugene Springfield Fire with 24 years of service. He writes Frontline Intelligence on doctrine, technology, and leadership in Fire & EMS.

If this landed, share it with someone in Fire & EMS who needs to hear it.

#FireService #EMS #AI #PublicSafetyLeadership #FireTech

Subscribe

From the floor, not the vendor booth. Two times a week.