In disaster situations, timely and well-informed decisions can make the difference between life and death. From coordinating emergency teams to directing resources in real time, disaster response requires accurate, fast, and consistent decision-making under high uncertainty. While humans remain central to these operations, AI offers the potential to augment their capabilities by processing vast amounts of data quickly and providing actionable insights.
A research team at Cranfield University, led by Professor Argyrios Zolotas, has developed a structured AI decision-making framework designed specifically for such high-stakes environments. This work, addresses the dual challenges of speed and reliability in disaster response, while also emphasizing ethical, transparent, and accountable AI use.
Dcruz, J. G., Zolotas, A., Greenwood, N. R., & Arana-Catania, M. (2025). Structured AI decision-making in disaster management. Scientific Reports, 15(1), 32093. https://doi.org/10.1038/s41598-025-15317-w
The framework organizes autonomous decision-making into multiple levels, incorporating Enabler agents and Decision Maker agents that process information from diverse sources, including satellite imagery, UAVs, and social media reports. By structuring the flow of decisions and making AI outputs traceable, the framework not only improves the quality of decisions but also provides a system that human operators can understand and trust.
In rigorous evaluation studies, the AI framework demonstrated 39% higher accuracy than human operators and 60% greater stability across multiple disaster scenarios. These improvements are especially significant in situations where human judgment may be inconsistent or delayed due to stress or incomplete information. The researchers also simulated scenarios including natural disasters such as floods, earthquakes, and urban emergencies, highlighting the framework’s adaptability across a wide range of critical situations.
Argyrios Zolotas, Professor of Autonomous Systems & Control at Cranfield University stated,
“Bringing AI into disaster response is not just about creating smarter algorithms; it’s about helping to facilitate faster, safer, and more resilient decision-making when lives and critical infrastructure are at risk. It’s vital that we go forward in a responsible way to ensure that AI use is ethical, transparent, and provides reliable outcomes. Our work in this study gives a valuable view of one way this can be achieved.”
Professor Zolotas emphasized the importance of responsible AI in disaster management: “Bringing AI into disaster response is not just about creating smarter algorithms. It is about facilitating faster, safer, and more resilient decision-making when lives and infrastructure are at stake. Ensuring that AI is ethical, transparent, and reliable is critical.”
The research also included human evaluation studies, where emergency management professionals assessed the AI’s outputs in realistic simulations. Participants reported that the structured framework made AI recommendations easier to interpret and integrate into operational decision-making, potentially bridging the gap between autonomous systems and human teams.
Beyond immediate accuracy, the framework’s event-driven processing reduces unnecessary computation and focuses AI resources on situations that require urgent attention. This efficiency is particularly relevant for emergency systems with limited bandwidth, computing power, or communication constraints. By prioritizing critical updates, the AI can respond to dynamic changes in disaster conditions without overwhelming human operators or infrastructure networks.
The Cranfield team views this framework as a foundational tool for integrating AI into real-world emergency systems. Potential applications include optimizing resource allocation in relief operations, coordinating autonomous drones for rapid damage assessment, and supporting decision-making for first responders in complex, fast-changing environments.
While the results are promising, the team notes challenges ahead. Real-world deployment will require robust integration with existing emergency management protocols, validation across diverse geographic regions, and consideration of evolving social, ethical, and technical requirements. Nevertheless, this research represents a meaningful step toward AI systems that are not only faster and more accurate than human operators but also responsible, interpretable, and trustworthy.
This work underscores a broader trend in AI research: moving beyond theoretical models to solutions that can assist humans in complex, high-stakes scenarios. As natural disasters and emergencies continue to pose global challenges, AI frameworks like the one developed at Cranfield University have the potential to transform how societies respond, making interventions more efficient, reliable, and ultimately, life-saving.
Adrian graduated with a Masters Degree (1st Class Honours) in Chemical Engineering from Chester University along with Harris. His master’s research aimed to develop a standardadised clean water oxygenation transfer procedure to test bubble diffusers that are currently used in the wastewater industry commercial market. He has also undergone placments in both US and China primarely focused within the R&D department and is an associate member of the Institute of Chemical Engineers (IChemE).
