How US emergency managers are keeping up with the AI revolution

The use of artificial intelligence in emergency management is rapidly moving from theory to practice in the United States, driven by a major philanthropic initiative that aims to bridge the gap between cutting-edge technology and the real needs of emergency managers on the ground.

At the heart of this work is the AIDE Initiative, a collaboration funded by the Markle Foundation and delivered with partners including Aspen Digital and RAND. I met former FEMA leaders Rich Serino and Jeremy Greenberg to find out more about this project which has been designed to look across the whole spectrum of emergency management and accelerate responsible, practical AI adoption.

It is a timely conversation because The Emergency Services Show takes place this month, and in the US September is National Preparedness Month. Both have been going for the last 20 years and have witnessed the ebb and flow of challenges facing our emergency services but the emergence of AI is set to be profound.

Emergency management explained

First, some language differences. Emergency management is not a phrase we use in the UK. We talk about emergency planning, resilience and the emergency services. When US practitioners talk about emergency management they mean something very broad that spans first response, preparedness, recovery, and the many organisations involved. Jeremy expands on this: “It’s not just public sector organisations as you’d recognise in the UK but public works, other state and local organisations, and then voluntary organisations, community and faith-based organisations.”

Rich is a senior advisor to the Markle Foundation and former deputy administrator at FEMA, the Federal Emergency Management Agency. He is keen to stress that the whole community approach is deliberate, and for me, it chimes with the phrase ‘whole of society’ which has been adopted by policy makers here in the UK to describe the need for everyone to be concerned about and contribute to a resilient society.

The AIDE Initiative is rooted in the concern that disaster risk and technological change are accelerating at a pace that may not be clear to all the organisations in the emergency management ecosystem.

“If you think about where we were as a world, as a community two years ago in the AI space, it is vastly different than where we are today. You have this one line arcing up of increasing disaster frequency and intensity and then this other line that’s intersecting where we see technology moving at a speed that we are all trying to make sense of. So those two points intersecting is really the focus behind the AIDE Initiative.”

Research findings

We speak after the publication of a report called AI for Disasters + Emergencies: A way forward. It’s an assessment of the way emergency management organisations have adopted AI including an analysis of available technologies along with a set of actions to support the advancement of responsible use of human-centered AI throughout the emergency planning lifecycle.

A core part of the work has been detailed research into the AI tools already on the market. RAND’s technical survey identified 1,179 AI-enabled products from 717 companies mapped against 45 emergency management task areas. These tools range far beyond familiar large language models and transcription services, and include predictive analytics, drones, computer vision and automation tools that support planning, grants, exercises and recovery.

But the research quickly showed an adoption gap between what the market offers and what emergency managers can realistically use. Greenberg sees that gap on both sides: technology companies often don’t fully understand what emergency managers need across preparedness, mitigation, response and recovery; at the same time, emergency managers rarely have the time or capacity to explore what is available.

Using himself as an example, he tells me that a year ago he would have defaulted to talking only about large language models; now, after talking to “a wide variety of people in tech, academic, research, and emergency management,” his understanding is far wider.

To help practitioners close that gap, the Initiative is developing an adoption playbook for emergency managers. This is not about picking winners in the marketplace, but about giving them a structured way to think about objectives, risks, procurement pathways and governance when they consider any AI tool.

They also need to know that what they are buying works for what they need and the challenge is knowing when to make the commitment to a tool when the technology is moving so fast. The AIDE Initiative can help with that, they are neutral and their intent is to get the best information to the right people at the right time, and take the opportunity to educate them.

AI literacy

Education and AI literacy are central goals for this work and Greenberg is clear that they want it to lead to an increase in effective use of the tools available to emergency managers. That includes AI usage and AI adoption so that it is used in responsible ways to help increase the capacity of emergency managers. He emphasises that AI should augment, not replace, core professional skills. The modern emergency manager must still train and exercise in the traditional way, while adopting approaches to AI that will see it pick up repetitive, predictable tasks such as note taking, drafting plans, or supporting grant applications.

I share my concern that early career staff may use AI tools without a solid foundation of knowledge and come up against problems as a result. Serino rejects the idea that AI could deskill the workforce. He argues that younger practitioners, in particular, will help drive innovation if they are supported properly. “AI isn’t going to take your job,” he cautions. “Somebody who knows AI is going to take your job.” That’s a phrase I’ve heard before but he is firm in his view, adding that the answer to how to use AI effectively in the world of emergency management is multigenerational learning, mentorship, where AI is embedded into career-long professional development.

Gathering a community

Looking ahead, the AIDE Initiative is moving from research into action. A 200-person cross-sector summit in Texas this month will bring together public sector emergency managers, private sector technology and infrastructure companies, and civil society organisations. From there, the team will launch a series of funded pilot projects from January, likely running for around a year, to test AI tools in real-world emergency management contexts.

Serino describes one of the biggest remaining challenges as connecting what he calls “cylinders of excellence.” By this he means, the many specialist organisations and sectors that must work together during crises. It seems that data sharing is as tricky to achieve in the US as it is here in the UK, with public sector organisations running systems in silos and mired in legacy debt that makes it hard to free up time to do the innovation and have clean data with which to do it.

We know that AI, used thoughtfully and collaboratively, can increase capacity and free up human attention for the most critical tasks but the approach is critical and needs to be based on trust and confidence that the outcome will enhance public safety. It looks like the US and UK have a lot to learn from each other as the challenges are similar and the work of this organisation is something we will be watching as they continue on their way.

For more conversations about AI and how it can help emergency planning, join us at The Emergency Services Show where Emergency Planning Society Chair Jeannie Barr will lead a discussion about dealing with AI slop when communicating during major incidents. The session takes place at 1510 on 16 September on the Protecting the Protectors stage. Register for your free ticket here.