AI and Climate Resilience Governance

Climate change is increasing the frequency and intensity of natural disasters, presenting complex policy challenges and non-linear risks that traditional modelling methods struggle to capture. This study investigates the potential of artificial intelligence (AI) to enhance climate resilience governance by leveraging its advanced analytical capabilities.

The research combines insights from an expert survey with a systematic literature review, highlighting AI’s promise in supporting climate resilience governance while identifying challenges such as data quality, model interpretability and ethical considerations.

The experts who responded to the survey cited AI’s ability to manage complex problems, process large datasets, identify patterns, predict trends and support decision-making as key advantages.

The authors – Sara Mehryar, Vahid Yazdanpanah and Jeffrey Tong – say these capabilities can be leveraged across the three stages of climate risk governance: risk assessment, policy appraisal, and implementation.

Currently, the use of AI methods is concentrated on the first stage – improving climate-related financial risk assessments. Of the literature reviewed, 64% focused on risk and resilience assessment, mainly in the fields of disaster management, flood and drought risk, and agriculture.

Machine learning techniques like random forests and neural networks are widely used in probabilistic forecasting, remote sensing and dynamic network analysis. These can aid in modelling climate hazards, forecasting extreme weather and assessing climate impacts.

In the risk assessment context, AI has primarily been applied to hazard and exposure evaluations, with less emphasis on the resilience and vulnerability components. The limited focus on community resilience and vulnerability is largely due to the scarcity of reliable socio-economic data, presenting an opportunity for development.

Around 29% of studies applied AI to policy analysis, mostly evaluating retrospective impacts. According to the authors, more research is needed to simulate and measure potential future policy impacts, including co-benefits and maladaptation consequences, before implementation.

Only 6% of studies considered how AI can be used to support the implementation of adaptation and resilience actions. Such studies emphasised consideration of how AI can improve communication and information sharing, resource allocation models, and post-disaster infrastructure resilience.

Despite AI’s potential, significant challenges remain, including data quality issues and ethical concerns like algorithmic bias and privacy.

Prioritising high-quality data collection and validation, enhancing AI model transparency, and ensuring a continued role for human judgement are key to ensuring their responsible use in decision-making processes. This is particularly important in the context of training data regarding how climate risks cascade through social systems and lower probability but extreme climate risks.

The study suggests future research should focus on developing AI tools for simulating complex long-term changes and evolving climate policies, improving human behaviour modelling, and conducting interdisciplinary research for AI-informed resilience governance.

The interpretability of complex AI models is another significant challenge put forward in the article, as even experts often struggle to fully comprehend and clearly interpret the outputs. According to the authors, this can impact the translation of model results into concrete actions and policies.

Finally, decision-makers must consider trade-offs between efficiency gains from AI analysis and the environmental impact caused by AI’s high computational resource requirements.