Climate-related evacuations consistently result in chaos. However, with the support of a National Science Foundation Faculty Early Career Development (CAREER) award, Xilei Zhao, Ph.D., an assistant professor in the University of Florida’s Department of Civil & Coastal Engineering, aims to utilize trustworthy AI to enhance emergency response modeling, training, and education. Her efforts are focused on improving evacuation procedures and reducing the uncertainty associated with climate disasters like hurricanes and wildfires.
“The findings from the research can be used right away to change how emergencies are handled. Imagine humans and AI working together in real-time during disasters,” Zhao said. “For instance, AI can help manage traffic during evacuations by using advanced behavioral models. This could completely transform how we handle evacuations, making them safer and more efficient.”
She describes the current gaps in existing behavioral models as bottlenecks for progress in emergency response situations. She explained her research, which starts May 1, will build a strong base for making AI models that predict human behavior during emergencies.
However, current research on these models is in the early stages, and it is not thoroughly considered trustworthy AI. To fill that gap, Zhao’s team is building theory-informed, high-accuracy AI-based behavioral models that are explainable and with minimized bias to inform decision-making in emergency management.
“To be trusted, an AI system must operate competently, interact appropriately with people, and meet pressing emergency management needs in an ethical manner. These models will be accurate, easy to understand, and won’t be unfairly influenced by biases. They’ll help people make important decisions during disasters,” Zhao explained. “Plus, they’ll train experts who can use these tools to prepare for and handle climate-related disasters better in the future.”
Her team is looking to transform emergency management by enabling real-time human-AI teaming, revolutionizing traffic management, and advancing equitable resource allocation during disasters.
Research findings will improve communications and evacuation procedures. This will impact public outreach campaigns, protective action training protocols, and warning strategies to better prepare households for future events.
Beyond improving training for emergency response, Zhao looks forward to engaging with younger generations to improve AI literacy. In the third and fourth years of this study, she aims to introduce AI concepts to students at various grade levels, enriching their understanding of this technology. Zhao is part of UF’s Scientist in Every Florida School (SEFS) program that connects scientists with K-12 schools across Florida through classroom visits. The program incorporates research findings into an innovative education activity for high school students, and it leverages the education outputs to facilitate research development. She will work with the SEFS’s program coordinator to host a two-day workshop at UF, inviting teachers to view the research findings, co-develop a curriculum, and select the case study.
This research will act as a baseline case study, which will improve emergency management and evacuations in the nation and eventually, throughout the world. In addition to this research, her long-term goal is to expand and adapt the framework to tackle various aspects of trustworthy AI and human behavior modeling across different hazards and beyond.
—
By Reba Liddy
ESSIE Marketing and Communications Specialist