Earlier this month, a meteorologist in Alabama alerted emergency management expert Laura Myers to a concerning trend: a local emergency manager had used an artificial intelligence chatbot to generate a graphic about Hurricane Isaias, but the resulting product contained meteorologically inaccurate information about a weather phenomenon known as a sting jet. While most emergency management agencies officially restrict generative AI use, experts agree the technology is already widely deployed across state and local disaster departments for various purposes, including communication with the public.
Emergency managers face a persistent challenge in crafting effective public alerts. Personnel at these agencies often struggle with communicating in accessible language, tend to use specialized jargon, and sometimes omit or include irrelevant details. Several documented cases illustrate the consequences of poor emergency communication. In 2016, neighboring California counties issued conflicting evacuation directives for the same wildfire. Australian officials’ use of the phrase “Watch and Act” in 2019 generated widespread confusion. Most notably, inadequate communication about a post-fire mudslide in Montecito, California, in 2017 contributed to 23 deaths, as residents fatigued from recent evacuations failed to leave during a subsequent threat.
Proponents argue AI systems could substantially improve emergency alerting. Large language models can rapidly process hundreds of thousands of historical alerts and research findings, then apply these insights to refine communications in seconds. Rob Dale, a former meteorologist now directing emergency management in Ingham County, Michigan, has experimented with AI-drafted practice alerts and expressed willingness to use the technology in actual emergencies, provided he retains final approval authority. Some researchers and startups contend the technology has matured sufficiently for deployment, with EM1, an emergency management software firm, claiming its AI model exceeds the capabilities of most human emergency managers.
The stakes remain high as climate-related disasters intensify. Trust between government officials and the public remains essential during crises, and poor communication can directly threaten lives. As hundreds of companies market AI tools for emergency management tasks ranging from damage assessment to resource coordination, the sector continues grappling with questions about accuracy, oversight, and the appropriate role of automation in public safety communications.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI