Perceptions, hopes, and concerns regarding the possibilities of artificial intelligence in weather warning contexts

Abstract

Artificial intelligence (AI) is increasingly used in disaster risk reduction, including early warning systems (EWS) for weather hazards. While AI promises faster data processing and improved forecast accuracy, concerns remain about automation bias, reduced human oversight, or accountability, and erosion of professional judgment. Despite rapid technological advances, the perceptions of the weather warning community remain underrepresented in current research. To address this, we conducted an Argumentative Delphi study with experts from the 2024 WMO HIWeather Final Conference. Participants assessed AI's impact on 13 key aspects of weather warnings – including quality, interpretability, accountability, and social bias – and shared hopes and concerns. Overall, participants expressed cautious optimism. AI is expected to improve the goodness of warnings, potentially cascading into broader dimensions of warning efficacy, public trust, and institutional responsibility. However, concerns include over-reliance on AI, erosion of human involvement, and challenges in maintaining a single authoritative voice in warning communication. Rather than viewing AI as replacement for human decision-making, it is seen as decision-support tool that augments professional expertise. Tailored warnings and multilingual communication emerged as promising areas for AI application, though issues of data bias and accessibility remain. Thus, ethical implementation is vital to ensure inclusiveness and alignment global disaster risk reduction goals. Finally, the introduction of AI touches the ‘professional core’ of weather warning as an occupation and prompts experts to define their evolving roles and core competencies in the face of technological advancements. Future research should explore how generative AI may reshape forecasting and the profession itself.