AI and Climate Risk

In the episode, Dale Hall and Carlos Orocha explain why conventional actuarial techniques struggle with climate risk. Traditional models rely on the assumption that future losses will resemble past experience, but climate variables are non‑stationary and driven by shifting weather patterns, policy changes, and technology adoption. This volatility makes historical loss data unreliable for predicting floods, wildfires, or heat waves. As a result, actuaries need tools that can capture dynamic, long‑term trends rather than static, repeatable events.

Adapting models now is critical for resilience. The conversation turns to how artificial intelligence, especially machine‑learning models, can fill that gap. By combining satellite imagery, topographic data, geolocation, and high‑frequency weather records, AI can evaluate individual buildings’ elevation, roof condition, surrounding vegetation, and resilience measures. This asset‑level insight replaces coarse proxies such as zip codes, producing far more granular hazard maps and refined severity estimates. Moreover, AI accelerates model updates and enables forward‑looking scenario analysis, giving insurers and reinsurers a clearer picture of emerging physical risks across diverse geographies.

Finally, the hosts highlight AI’s impact on scenario and stress testing, as well as transition risk assessment. Machine‑learning algorithms can link climate, economic, and portfolio data to simulate numerous physical and policy pathways, exposing concentration and tail‑risk exposures that traditional deterministic scenarios miss. Yet Carlos warns that AI should augment, not replace, actuarial judgment; data quality, model governance, and explainability remain essential. He advises professionals to blend rigorous actuarial training with climate science and responsible innovation, ensuring AI‑driven insights are transparent, resilient, and actionable for the evolving risk landscape.