AI Integration and Environmental Crisis Management: Policy Capacity as a Bridge — What Can South Korea Tell Us?

This paper develops a theoretical framework for understanding how central governments integrate artificial intelligence (AI) into the management of environmental crises, arguing that policy capacity constitutes a set of enabling conditions for such integration. While AI is increasingly embedded in public-sector decision processes, its adoption is neither automatic nor just technological. Governments must reconfigure their logic and structures to make AI outputs actionable, legitimate, and embedded within crisis governance. This study demonstrates that policy capacity - analytical, operational, and political - functions as the bridging mechanism through which this reconfiguration occurs. Building on crisis management scholarship, the contribution investigates each dimension of policy capacity in the phases of the crisis management cycle: sense-making, decision-making, meaning-making, accounting, and learning. Analytical capacity supports the detection and interpretation of emerging threats through knowledge acquisition, modeling, and evidence-oriented organizational culture. Operational capacity underpins decision-making and coordination, aligning resources, responsibilities, and procedures to translate analytical insights into coherent action. Political capacity anchors meaning-making and accountability processes, shaping legitimacy, trust, and the communicative authority required to sustain crisis responses. Learning encompasses all dimensions, reinforcing and reshaping governments’ capacities. The paper argues that these same dimensions also constitute the enabling conditions for AI integration. Analytical capacity provides the expertise, infrastructure, and epistemic foundations necessary for data collection, processing, and interpretation - the cognitive prerequisites for AI systems and their outputs to contribute to crisis governance. Operational capacity enables interoperability, ensuring that AI outputs can circulate across administrative boundaries and become embedded in shared routines, responsibilities, and decision structures. Political capacity supports the legitimacy of AI-driven governance by addressing transparency, accountability, and public trust, mitigating algorithmic aversion, and affirming the political acceptability of using AI as a digital policy tool. Together, these dimensions reveal that AI integration and then adoption require profound institutional adjustments. This framework demonstrates theoretically that policy capacity is not merely a background administrative feature but an active bridge connecting AI integration in governments and their environmental crisis management. The empirical component of the contribution applies this framework to South Korea, focusing specifically on policy analytical capacity. The paper applies and operationalizes the policy analytical capacity to examine how governments integrate AI outputs during the crisis sense-making phase. It is based on a qualitative investigation of a case study: the adoption of an AI Long Short-Term Memory model for flood prediction by South Korea’s Ministry of Environment. The study employs semi-structured interviews with engineers involved in the project, policymakers, and experts to explore and assess project functioning, data governance practices, technical capabilities and attitudes of officials, organizational commitment to evidence, etc. Overall, the paper contributes a conceptual lens for understanding and (partially) assessing the administrative enabling conditions and remaining gaps – at individual, organizational, and systemic levels – for the integration of AI outputs in climate-related crisis management. The paper provides both a theoretical contribution to the literature on AI and crisis governance and empirical evidence on practices that are reshaping contemporary governance.