Tag
Article examines the risks and growing use of generative AI in drafting emergency alerts and public disaster communications, highlighting concerns over accuracy and trust as agencies adopt AI tools despite incomplete policies.
WIRED/Grist reports on the growing use of generative AI tools by emergency managers to craft public alerts, highlighting risks of inaccurate outputs and unprepared agencies as climate disasters intensify.
New satellite images from Copernicus and NASA reveal the destruction caused by the June 24 earthquakes in Venezuela, aiding rescue efforts and damage assessment.
This paper introduces PanicCognitivePath (PCP), a framework that combines cognitive graphs and LLMs to predict individual panic emotional arousal timing during emergencies, grounding prediction in appraisal emotion theory and demonstrating improved accuracy on the Hurricane Sandy dataset.
This paper introduces WLDS, a large-model-driven system for simulating and deducing emergency instances by leveraging controllable randomness and cross-domain knowledge. It presents the Emergency Instances Deduction (EID) benchmark and demonstrates high-fidelity simulation capabilities across multiple domains.