AI Council research with SAS identifies 16 potential applications and selects catastrophe exposure intelligence as the first pilot
Understanding complex risks such as exposure to natural catastrophes, claims litigation and emerging liabilities often requires insights beyond those available to any single insurer. New research from the AI Council, a joint initiative of the Insurance Information Institute (Triple-I) and The Institutes RiskStream Collaborative, developed with technology collaborator SAS, examines how synthetic data could support broader industrywide analysis while protecting sensitive policyholder and proprietary information.
The report, Synthetic Data in Insurance: Using Synthetic Data to Unlock Industry-Wide Intelligence, identified 16 potential applications for synthetic data, with catastrophe exposure intelligence selected as the first use case for a planned pilot project.
“The risks facing insurers increasingly extend beyond the experience of any single company,” said Sean Kevelighan, CEO, Triple-I. “By using synthetic data to facilitate privacy-focused contributory data solutions, we can strengthen our understanding of catastrophe exposures and emerging threats. Better information supports better risk management, stronger resilience and more informed decisions that benefit the communities we serve.”
Synthetic data is artificially generated information designed to reflect patterns and characteristics of real-world data without directly reproducing individual records. It offers a potential approach to analyzing broader risk patterns without requiring participating insurers to disclose original customer records or proprietary business information.
“One of the industry’s longstanding challenges to improving market dynamics for consumers is understanding claims behavior and exposures while protecting sensitive underlying information,” said Patrick Schmid, Ph.D., who oversees AI Council efforts, including research at Triple-I and solution development at RiskStream. "Synthetic data lets companies maintain their original records under their own control while securely transferring only what the industry needs to learn for an industry-wide perspective. If this approach works for catastrophe exposures, it could open the door for aggregated market insight on fraud, litigation and other risks, to the benefit of the entire industry and the consumers we serve."
The research identifies potential applications beyond catastrophe exposure, including insurance fraud detection, litigation trends, cyber risk and emerging liability exposures.
Catastrophe Exposure Selected for Initial Pilot
Individual insurers maintain detailed information about the properties they insure, but no single company has a complete picture of the industry’s combined exposure to hurricanes, wildfires and other natural disasters. More comprehensive exposure intelligence could strengthen catastrophe risk assessments and help insurers, researchers and other stakeholders better understand the potential financial impact of major disasters.
The report explores the impacts of Hurricane Ian, which struck Florida in September 2022, as an example of the challenges associated with catastrophe loss estimation. Within 10 days of the Category 4 major hurricane, five catastrophe modeling organizations issued insured loss estimates ranging from $30 billion to $74 billion. Differences in modeling assumptions and the types of losses included contributed to the variation, but limited visibility into insured property exposures presented additional challenges.
The planned pilot will examine whether appropriately governed synthetic datasets can improve understanding of geographic exposure concentrations and potential catastrophe losses without disclosing participating insurers’ original claim data.
Under the proposed approach, participating insurers would generate synthetic datasets within their own technology environments, retaining control of their original information. Certain geographic characteristics would be preserved to support catastrophe modeling, while sensitive property and financial information would be artificially generated. The pilot will evaluate the resulting datasets for analytical accuracy, usefulness and privacy protections.
SAS is providing SAS® Data Maker, the software selected for the pilot, to support the creation and evaluation of synthetic datasets.
“Every insurer sees just a piece of the broader market risk picture. One challenge is connecting those pieces when the data behind them is too sensitive to share,” said Stu Bradley, senior vice president of risk, fraud and compliance solutions at SAS. “This collaboration presents a unique opportunity to put synthetic data to the test on one of the toughest challenges facing the industry today. If insurers can see where catastrophe exposure is concentrated across the market and more accurately model that risk and potential loss, they can make better decisions and take action as those risks evolve.”
Organizations interested in participating can contact The Institutes RiskStream Collaborative at RiskStreamDL@theinstitutes.org.
About the AI Council
The AI Council is a joint insurance industry initiative comprising the AI Policy Council, led by the Insurance Information Institute (Triple-I), and the AI Solutions Council, led by The Institutes RiskStream Collaborative. Together, the councils bring research, education, policy expertise and practical technology development to the responsible advancement of artificial intelligence in insurance. The AI Council involves more than 35 organizations across the insurance ecosystem.
About the Insurance Information Institute (Triple-I)
Since 1960, the Insurance Information Institute (Triple-I®) has been the trusted voice of risk and insurance, delivering unique, data-driven insights to educate, elevate and connect consumers, industry professionals, policymakers, and the media. An affiliate of The Institutes, Triple-I represents a diverse membership accounting for nearly 50% of all U.S. property/casualty premiums written. Our members include mutual and stock companies, personal and commercial lines, primary insurers and reinsurers – serving regional, national and global markets.
Triple-I is a registered trademark of the Insurance Information Institute. All rights reserved.
About The Institutes RiskStream Collaborative
The Institutes RiskStream Collaborative aims to create an ecosystem within the risk management and insurance industry that leverages emerging technology platforms to design/build solutions which aim to streamline the flow and verification of data in multiparty business processes. The goal is to lower operating costs, drive efficiency from improved processes, and enhance the customer experience.
About SAS
SAS is a global leader in data and AI, helping organizations make confident decisions with AI they can trust. For decades, SAS has set the standard for delivering software that drives meaningful impact, incorporating deep industry expertise, transparency and governance. SAS gives you THE POWER TO KNOW®.
View source version on businesswire.com: https://www.businesswire.com/news/home/20261119519743/en/
“By using synthetic data to facilitate privacy-focused contributory data solutions, we can strengthen our understanding of catastrophe exposures and emerging threats."
Contacts
Media Contact:
Media@iii.org
If you believe this article contains misleading, harmful, or spam content, please let us know.
Report this article