In a groundbreaking study led by the University at Buffalo, ChatGPT has been successfully trained to extract location information from social media posts during natural disasters. The study, published in the International Journal of Geographical Information Science, represents a significant leap forward in the use of AI in emergency response scenarios. The research, which also involved collaborators from the University of Georgia, Stanford University, and Google, aims to develop AI systems that can process social media data automatically, aiding emergency services without requiring AI technology expertise.
Transforming AI Models to Aid First Responders
The research team trained geoknowledge-guided GPT models that demonstrated a marked improvement in location recognition from tweets. The enhanced models outperformed their default counterparts by a whopping 76%, and proved 40% more effective than named entity recognition (NER) tools. This advancement was achieved by training the AI with specific prompts containing real-world examples of location descriptions from tweets sent during Hurricane Harvey. The GPT-3 and GPT-4 models learned to identify various types of locations such as addresses, intersections, and landmarks, a crucial element during emergency response.
AI: The Future of Emergency Response
The research commenced in early 2022 and has continued with newer GPT models released in late 2022 and early 2023. The researchers aim to further refine the system to discern between relevant and irrelevant posts during a disaster. This initiative could revolutionize emergency response, allowing first responders to reach victims more efficiently and saving more lives in the process. The study's findings underline the transformative potential of AI, not just in the realm of emergency services but across multiple sectors.
AI's Expanding Footprint in Disaster Management
Elsewhere, AI is also being used to navigate drones in adverse weather conditions or low visibility during disaster response and offshore wind turbine inspections. A solution leveraging passive millimeter wave RFID tags to help drones identify ground locations has been released by the NTT Group. Moreover, the hospitality industry is integrating AI monitoring systems to meet evolving customer safety and hygiene requirements, highlighting AI's potential to revolutionize sustainable development and foster global ecological resilience.