In the wake of worldwide natural catastrophes, vulnerable countries are turning to advanced machine learning as a resource in disaster relief.
Emergency Search and Rescue Efforts in Turkey. State Emergency Service of Ukraine. CC BY 4.0.
On Nov. 6, 2024, an inaugural meeting at the Barcelona Supercomputing Center for the Global Initiative on Resilience to Natural Hazards through AI Solutions was held to address the role of artificial intelligence in natural disaster management. The initiative, a collaborative effort involving many environmental protection UN agencies, seeks to utilize AI and machine learning to build resilience to natural hazards while reducing their risk of happening. This increasing need for precaution follows a dramatic rise in and worsening impact of natural disasters around the world. As the International Disaster Database demonstrates, there has been a worldwide increase in disaster events, from about 100 per year in the 1970s to about 400 per year as of 2023. With more than 4 billion people living in urban areas, threats posed by extreme weather events are becoming more significant, as is finding a way to handle them.
The Global Initiative responds to this by outlining how AI can enact change. Machine learning, in particular, is a branch of AI in which computers and machines imitate how humans learn and perform tasks by gaining experience and receiving data. This form of technology has been used for years to forecast weather models, based on patterns of weather from past data and physics investigations, and has only advanced in scope and application. Countries have begun taking advantage of this progress; for example, the Philippines’ Department of Science and Technology (DOST) and the South Pacific island of Tuvalu have made deals with AI meteorology company, Atmo, to use their forecasting technology to predict weather. Addressing the environmental and economic damage caused by climate change, such as rising sea levels and tropical cyclones, Atmo’s technology provides weather forecast models and resolutions formed by graphical processing units created by numerical weather predictions and AI deep neural networks.
In addition to prevention systems, AI is also being used in disaster relief efforts. In 2023, southern Turkey and northern Syria experienced a magnitude 7.8 earthquake, immediately followed by another one of magnitude 7.5. Along with the damage and death toll initially caused by this catastrophic event, many survivors were left at risk due to dangerous destruction and cold weather conditions, as well as a lack of necessary resources to aid those affected. However, efforts to respond to this devastation were enhanced by AI. Programs like xView2 collaborated with research partners such as Microsoft and the University of California to combine machine learning with satellite imagery in identifying infrastructure damage and categorizing its severity at a very fast rate, enabling first responders to assess the situation in many areas and coordinate recovery efforts.
The Global Initiative aims to advance such natural disaster management techniques. To do this, it also intends to form guidelines on how governments can responsibly use such technology, including addressing its limitations. Because AI and machine learning enhance performance through the data they receive, any lack of data or incorrect information may lead to inaccurate results. However, collaborative efforts to preemptively address these concerns are a step toward using AI frameworks as a trustworthy tool in natural disaster management.
Julia Kelley
Julia is a recent graduate from UC San Diego majoring in Sociocultural Anthropology with a minor in Art History. She is passionate about cultural studies and social justice, and one day hopes to obtain a postgraduate degree expanding on these subjects. In her free time, she enjoys reading, traveling, and spending time with her friends and family.