Hype versus Historical Continuity: Situating the Rise of AI in Climate and Disaster Risk Modeling

Research Background and Issues

• Identified Problems or Challenges: The author points out that disaster risk modeling originally emerged from the insurance industry. However, with advancements in technology and data, these models have been widely applied in public sector decision-making. The author questions whether these models adequately reflect social complexities and the unequal impacts of disasters on vulnerable groups. Furthermore, while the introduction of artificial intelligence (AI) promises increased efficiency and accuracy, it may exacerbate issues such as bias, lack of transparency, and inequitable resource distribution.

• Significance: Climate change has led to an increase in the frequency of extreme weather events, making accurate disaster risk assessment increasingly critical for reducing harm and promoting equity. If existing models and their algorithms continue to overlook social and localized factors, they may result in unfair disaster management policies.

• Research Motivation and Related Work: Inspired by studies on "risk governance" and historical technological modeling, the author seeks to explore, from a historical perspective, how disaster risk models have been profoundly influenced by insurance logic. The research further examines whether the application of AI in this field can truly promote equitable policies rather than perpetuate existing systemic issues.

Solutions

• Proposed Methods or Solutions: Using the conceptual framework of "insurance logic," the author analyzes the continuities in the evolution of disaster risk modeling, particularly how AI technology reshapes the "calculative," "financialized," "collective," and "managerial" aspects of existing disaster risk models. Through document analysis, the study reveals the technical and social origins of risk models and their implications for current AI applications.

• Innovative Contributions: The introduction of the "insurance logic" conceptual framework provides a systematic analysis of disaster risk models and their evolution from four dimensions: calculative, financialized, collective, and managerial. This framework not only examines the technology itself but also elucidates its embedded relationships with social, economic, and political contexts, offering a critical perspective for disaster governance in the public sector.

• Implementation Steps and Key Technologies:

  1. Develop an analytical framework defining the four logics (calculative, financialized, collective, managerial).
  2. Conduct document analysis to study the technical standards, policy guidelines, and enduring insurance logic in related texts.
  3. Investigate how AI perpetuates these logics and evaluate its practical applications in disaster risk modeling.

Research Findings

• Specific Findings: The author discovers that although disaster risk models have expanded from the insurance industry to public decision-making, the core logics of these models have not fundamentally changed due to technological upgrades or new use cases. AI is primarily used to enhance the precision and efficiency of existing technologies rather than address structural social issues. The author identifies four enduring logics within risk models:

  1. Calculative: Risks are quantified into probabilities and loss data, neglecting social and cultural complexities.
  2. Financialized: Models often serve financial markets or insurance needs rather than actual disaster risk reduction.
  3. Collective: Models tend to generalize, failing to account for individual or localized differences.
  4. Managerial: Risk governance is dominated by remote experts, prioritizing scientific precision in data and models.

• Comparison with Existing Solutions and Advantages: Unlike other studies on AI in disaster risk management, the author provides a historical and structural perspective, uncovering how risk modeling entrenches existing practices and perpetuates social biases. These findings challenge the common narrative that AI is a "disruptive" or "breakthrough" technology.

• Experimental or Evaluation Results: Through historical case analysis, the study finds that AI technology primarily extends the existing logics of disaster risk models, serving to validate or reinforce established frameworks rather than initiating a paradigm shift.

• Limitations and Future Directions:

  • Limitations: The research relies on historical case studies and document analysis, which may not fully capture the regional and methodological diversity of disaster risk assessments. Additionally, the proprietary nature of some data and models limits detailed analysis.
  • Future Directions: Explore the true innovative potential of AI in disaster risk modeling, conduct in-depth evaluations of participatory modeling approaches, and investigate how existing technological and policy structures can be transformed to achieve more equitable disaster governance.

Through this analysis, the author not only maintains a critical stance on the social applications of AI but also provides methodological tools for scholars and practitioners to uncover potential issues and design alternative solutions.