The Black Mirror of AI-driven vulnerability, hazard, and risk research

As artificial intelligence (AI) is increasingly integrated into disaster risk reduction/management (DRR/M), critical scrutiny remains essential to avert unintended consequences that may compound existing challenges. This study aims to explore the use of AI in hazard, vulnerability, and risk assessments through the analytical lens of the Black Mirror conceptual framework, focusing on (1) the critical shortcomings of these integrations and (2) the potential dangers of translating faulty research outputs into DRR/DRM policy and practices. A semi-systematic review of the literature underpins these identified limitations. While observable in AI applications, the presented aspects are not confined to them. To address the identified concerns, we outline 5 rules of thumb that create a roadmap for improved AI development and integration. We contend that progress in DRR and DRM will be determined less by the sophistication of AI models and more by the intellectual and ethical rigor we apply in their development.