How insurers are using AI to find flood coverage opportunities

John Dickson, Rajiv Matta and Mark Pinkerton
John Dickson, president of AonEdge; Rajiv Matta, chief innovation officer, Lilypad Insurance; and Mark Pinkerton, CTO of Previsico.

Takeaways:

Processing Content
  • Insurers use AI to address flood insurance gaps
  • Aon using AI to price risk and support customer decision-making
  • Platforms, insurers apply AI to evaluate forecasting, analyze risks

Private insurers are developing AI technology to enable them to insure more properties when the National Flood Insurance Program's (NFIP) latest temporary funding ends in September.

Severe floods in recent years, like Ruidoso, N.M., last July and Asheville, N.C., after Hurricane Helene in September 2024, happened in places where flood insurance was not mandatory for home mortgages. This presents a real risk for many more people than in years past, said John Dickson, president of AonEdge, the flood insurance division of global risk mitigation company Aon.

"We want to close that coverage gap. There's a great application for AI. I see a role for AI in two categories, how insurers can use AI to more efficiently price their business — that's the back-office AI solution," Dickson said. "The customer-facing AI aspect is, how can AI be used to help people make better decisions? That's part of the challenge."

Flood maps can be 10 years out of date or more, based on weather patterns that have changed. This is a gap that AI could address, Dickson said. 

"AI can help an individual understand their specific risk vis-a-vis today's weather events, with the most current data and information out there, and help that individual make the best decision for that individual's needs," he said. "Not only is AI useful for locating updated data, it is also an outstanding resource for assimilating, organizing and thinking about new ways to deploy the data."

Insurers, brokers and reinsurers are all using AI to price flood risks, to measure appetite for these risks and to see risk accumulation, according to Dickson. Aon specifically uses AI and advanced analytics to break down large amounts of data relevant to flood risk, he explained.

"For flood, AI and advanced analytics help us ingest large amounts of hazard, exposure and loss data so we can give clients a more granular view of where their accumulations sit, how different scenarios might impact them, and what that means for their insurance strategies," Dickson said.

With more Americans moving away from flood risks, using AI to find less risky properties could give insurers more opportunities to write new business.

Using AI to assess risk is not new, with large language models being used by insurers to gauge catastrophe risk, Dickson notes. Aon recently began using AI to communicate risks to its clients to help them make coverage decisions. Overall, Aon is investing $1 billion in technology and analytics over a three-year period, but does not disclose what portion of that is specifically for AI, according to Dickson.

Previsico, a U.K.-based flood intelligence platform, uses AI to validate its forecasts by comparing them to evidence from ensuing weather events, according to Mark Pinkerton, the company's chief technology officer.

"We are now deploying AI to augment our proven hydrological models, expand into new flood types and turn our probabilistic forecasts into clear, actionable decisions for customers," Pinkerton told Digital Insurance in response to emailed questions.

Lilypad Insurance, a specialty insurer focusing on coastal home insurance, applies AI to NFIP claims data for more accurate analysis of flood risk, according to Rajiv Matta, chief innovation officer at the company.

"It's our job as the industry to keep building these custom products to fill the protection gap," he said. "AI is really helping with that, from a product design perspective, from an analytics perspective, from a technology perspective. It used to take a year to get a product from an idea to finish. That cycle of product development has been squeezed to about three months now, because you can build a system very quickly to code by an issue of policy."


For reprint and licensing requests for this article, click here.
Flood insurance Artificial Intelligence Property and casualty insurance Insurtech
MORE FROM DIGITAL INSURANCE
Load More