When a crisis hits the public directly, demand for fast and personal information often grows faster than a call centre or service desk can keep up with. AI chatbots can help, provided they do not only supply facts but also acknowledge what people are going through at that moment. This article covers what research says about person-centred AI communication with the public, where chatbots add value in practice, and where the line to human contact lies.
Why people need more than facts during a crisis
During a crisis, people mainly look for practical information: is my street safe, do I have to evacuate, where can I go. That need is real and urgent, and a good chatbot has to answer it quickly and correctly. At the same time something at least as important is going on: people living through a crisis often experience fear, uncertainty or grief. Those emotions do not disappear because a chatbot gets the facts right.
Research from 2025 into government chatbots during natural disasters shows this genuinely matters. Chatbots that communicate in a "verbally person-centred" way, acknowledging people's emotions and offering validation instead of only listing dry facts, significantly reduce emotional distress. That same person-centred approach also raises satisfaction with the service. How a chatbot says something turns out to matter at least as much as what it says.
For municipalities and safety regions that is a relevant lesson. A chatbot that only lists instructions is technically doing its job, but misses a chance to reassure people at the moment they need it most. A sentence acknowledging the person's situation, before the instruction follows, costs a chatbot nothing extra. For the recipient it can make the difference between feeling heard and feeling like a number in a system.
Practical use: a 24/7 information hub that stays consistent
Besides the tone a chatbot uses, there is the practical question of what a chatbot can actually deliver during a crisis. Two properties stand out in practice.
The first is the role of a 24/7 information hub. During a crisis the situation sometimes changes hour by hour, while phone lines and service desks cannot be staffed day and night. A chatbot that is continuously available can give people current information at any moment: the latest state of affairs, which roads are closed and which facilities are open. That relieves the crisis organisation exactly when demand for information is at its highest, while people do not have to go to bed with an unanswered, urgent question.
The second is consistency. Where different staff on the phone may answer a question slightly differently, a chatbot gives every person the same current information at any moment. For someone who has to decide what to do in the middle of a crisis, that consistency is as valuable as speed: contradictory answers undermine trust faster than a chatbot simply repeating the correct, current answer.
The line: when a chatbot must hand over to a person
However valuable a chatbot can be, there are clear limits to what automated communication should absorb. The first and most obvious limit is acute distress: when someone is in immediate danger or needs urgent help, a chatbot must refer without detour to the emergency number or the right service. No chatbot may cost time here that a person could have saved directly.
A second limit lies with complex individual situations: questions that do not fit a script, where personal circumstances call for tailored judgement a chatbot cannot provide. Referring to a person is then not a failure of the chatbot, but exactly how it should work.
A third limit, connecting to the research finding on person-centred communication, lies with clear signs of emotional crisis in the person themselves: panic, despair or signals pointing to acute distress. Here too: validation and acknowledgement by a chatbot are valuable, but they do not replace human contact at the moment someone needs it most. A well-designed chatbot recognises such signals and actively hands over, rather than trying to finish the conversation inside its own system.
What this means for organisations considering a chatbot
For municipalities and safety regions considering a chatbot for public communication during crises, a few clear points follow from the above. Set the chatbot up not only to convey facts correctly, but also for the tone in which it does so: validating and acknowledging what people are going through is not decoration but a demonstrably effective part of good crisis communication. Also make sure the chatbot genuinely gives current, consistent information updated at the moment of the question, rather than stale answers people could already have found on a general website.
Just as important is designing clear handover points to people. A chatbot that takes public communication during a crisis seriously is a chatbot that knows when to stop and refer. That calls for scenarios thought through in advance: which signals point to acute distress, which questions are too complex to automate, and how quickly the chatbot then hands over to a member of staff. Getting that right in advance draws the best from both worlds: the speed and availability of AI, and the human attention people need at the hardest moments.
Human in the loop, also outside the public chatbot
CrisisMaatje is designed as AI support on the side of the crisis organisation itself, not as a public-facing chatbot. But the underlying human-in-the-loop philosophy, where AI supports and a person keeps control at the moments that demand it, is the same.
Sources
- Taylor & Francis: Can AI Chatbots Support Me as Human Agents? (2025).
- M2SYS: 24/7 real-time AI chatbots for citizen crisis support.
About this article
Published by CrisisRadar B.V.
This article has been checked against primary sources: Dutch crisis management practice, European regulation and academic literature. The sources used are listed above, with dates. Spotted something that is wrong? Tell us; we will correct it and record the change. Read how we work on Editorial policy.
Try it yourself with CrisisMaatje
CrisisMaatje connects the AI you already use — Claude, ChatGPT or Copilot — to the CrisisRadar platform. You ask for help in plain language, your AI drafts a message and records the facts, the timeline and the statutory reporting deadlines in a crisis file. You review it and decide what goes out. Free, no credit card.