US Says Natural Hazard Models Need AI, Despite Challenges

Leveraging private market expertise is one solution the US government is mulling.

US Says Natural Hazard Models Need AI, Despite Challenges
Photo by Steve Johnson / Unsplash

Natural hazard models will see huge benefits from the rise of Artificial Intelligence (AI) but in order to compete the United States needs to address some fundamental issues around its development, including often chaotic and competing research efforts and a lack of confidence in the underlying data.

“Computer modeling is a critical tool for natural disaster forecasting,” the report by the Government Accountability Office (GAO) released earlier this month states, adding that the incorporating AI into hazard models remain largely in the research and development phase and “continued testing in operational settings will be needed to further mature the field.”

The benefits of AI in natural hazard models were various, the report states, including reducing forecast times, increasing accuracy and reducing uncertainty of model output.

The report focuses on the application of machine learning in natural hazard modeling, discussing its current and emerging uses for severe storms, hurricanes, floods, and wildfires, along with the potential advantages.

The report did not review the use of hazard and catastrophe models in the private market, such as in the insurance industry.

However, the challenges to the US advancing to the next generation of hazard models was significant.

A few of the challenges the GAO listed include:

  • Data limitations in some regions of the US, like rural areas that lack significant weather observations.
  • A “lack of trust and understanding of the algorithms” that make users hesitant to employ machine learning models.
  • The AI industries’ unwillingness to collaborate with forecasters to “interact with researchers and convey their needs.

Both funding and workforce resource challenges, including huge upfront AI development costs and private market players that “do not fully understand the data and phenomena they are modeling.”

Workforce and resource gaps also create challenges. For example, the upfront costs to develop and run machine learning models are high, and some companies working on these models do not fully understand the data and phenomena they are modeling, according to academic researchers.

The GAO presented policy options for government agencies to take to address the issue, including fixing shortfalls in data sharing and collection, education and training reforms and expanding public-private partnerships (PPPs) to expand resources and leverage private market expertise.

Great! You’ve successfully signed up.

Welcome back! You've successfully signed in.

You've successfully subscribed to Risk Market News.

Success! Check your email for magic link to sign-in.

Success! Your billing info has been updated.

Your billing was not updated.