Year: 2025
Type: article
Abstract: Large Language Models (LLMs) have emerged as transformative tools in crisis management and disaster response, offering real-time decision support, multilingual communication and rapid data synthesis for emergency responders and affected communities.This review examines the evolving landscape of LLM-based interventions in disaster scenarios, analysing their role in early warning systems, situational awareness, misinformation mitigation and humanitarian aid coordination.We provide a systematic evaluation of state-of-the-art models, including GPT-4, Claude and Mistral, highlighting their capabilities for processing unstructured crisis data from social media, satellite imagery and sensor networks.Additionally, we explore applications in emergency call triage, dynamic resource allocation and the automation of crisis communications to bridge information gaps in high-stakes environments.Although LLMs offer significant advantages, challenges persist in their reliability, bias and interpretability, necessitating robust fine-tuning, domain adaptation and human-in-the-loop strategies.We also discuss the ethical considerations surrounding the deployment of AI-driven crisis response tools, emphasising the importance of transparency, accountability and equitable access.This review presents recent advancements and case studies to highlight key research directions for enhancing the robustness, generalisation and real-world applicability of LLMs in humanitarian contexts.Interdisciplinary collaboration among AI researchers, policymakers and first responders plays a crucial role in ensuring that these technologies are effectively integrated into disaster management strategies.Leveraging LLMs as a force for social good requires ongoing refinement, ethical oversight and a commitment to developing AI systems that support equitable and efficient crisis response while mitigating the impact of global crises.
Authors: Oluwatimilehin Odubola, Tewogbade Shakir Adeyemi, Olaonipekun Olaitan Olajuwon, Nwaamaka Pearl Iduwe, Aniema Aniekan Inyang, Taiwo Odubola
Institutions: Microsystems (United Kingdom), University of Worcester
Language: English
Subfield: Sociology and Political Science
Field: Social Sciences
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W4408095418