AI alert tools spread as Hurricane Isaias graphic misfires

The Alabama episode sits inside a wider pattern the experts describe: emergency managers are human, and humans under pressure write alerts that use jargon, omit context, or bury the essential instruction. Jeannette Sutton of the University of Albany notes that agencies are limited in resources and people, so public communication gets attention only after something goes wrong. That gap is what makes generative AI attractive, because a large language model can read hundreds of thousands of alerts in minutes and apply research on what works.

Examples show the stakes are not abstract. In 2016, adjacent California counties issued conflicting "voluntary" and "mandatory" evacuation orders for the same wildfire. In Australia in 2019, the phrase "Watch and Act" caused what the government later called "significant confusion." After the 2017 Thomas Fire, a mudslide in Montecito killed 23 people and injured more than 160 others as Santa Barbara County struggled to communicate risk to residents already experiencing what Sutton calls "warning fatigue."

Against that record, experts are not simply bullish. Ginny Katz of HazAdapt warns that AI-generated content does not feel authentic, and members of the public skeptical of AI or government may spurn agencies that publish it. Sarah Miller of Pierce College says official adoption is low but unofficial use is real, with staff using personal or free accounts. The technology, the experts say, is only as capable as the person using it, and it is too early to tell how AI-generated content affects public trust.