Takeaways:
- Insurers should set guidelines for using AI with aerial imagery
- Regulations can vary widely
- AI can widen a transparency gap in use of images
AI increasingly plays a role in insurers' use of aerial images to determine home property coverage.
States that do regulate the use of aerial images tend to do so
As carriers adopt AI systems for aerial imagery, they are setting guidelines for its use, according to Renee Belgarde, executive vice president at Amwins Insurance Brokerage. "It's more about flagging," she said. Using AI to then determine what is in obscured parts of a property shown in an image is problematic, she explained.
"AI should be used as a guide to what is good or bad on a property, but let's ask some questions about it," Belgarde said. "Has the roof recently been done, or is that debris that's over to the left of this building still there? Or what is it? Maybe we're seeing it as debris, but it's actually maybe the top of a tree."
Applying AI to aerial imagery decisions creates more risk for insurers, advisors and agents say.
"Many states have a large gap: clear notification requirements regarding aerials are widely absent, resulting in homeowners often being unaware that such data has driven their decisions at policy levels. It is also difficult for them to dispute or contextualize any negative decision because of this lack of transparency," stated Brad Spurgeon, CEO of Brad Spurgeon Insurance Agency in Texas City, Texas, in a written response. "The gap in transparency is likely to widen as AI processing comes more and more to the forefront of image analysis before regulatory measures for public confidence can be put in place."
As with other applications of AI, insurers are still responsible for the result of AI-supported decisions, according to Franklin Manchester, principal global insurance advisor at SAS.
"As an underwriting professional who has reviewed tens of thousands of property photos, I would caution: Even a trained human can make mistakes," he stated in a written response. "Any insurer utilizing a third-party service to provide underwriting criteria into their processes will be held accountable for that data's use – right or wrong. If you're allowing the model to dictate those decisions, you are exposing your enterprise to significant risk."
The NAIC's model bulletin on AI, issued in 2023, and its subsequent plan for an AI Systems Evaluation Tool for regulatory inquiries, touch on insurers' AI applications using aerial images, according to Paige Waters, partner at Troutman Pepper Locke. "They want to understand how AI is used, to make sure that it's not being used in ways violating trade practice statutes, that it's not being used as a proxy for some unlawful bias," she said.
