What that means for your insurance

When you apply for a policy or file a claim, artificial intelligence is likely at work in the background, helping your insurance company respond.

Many insurers already use AI to help assess risk, detect fraud, and process claims. Industry experts say that use will expand.

Eighty-eight percent of car insurance companies and 70% of home insurers reported that they use or plan to use or explore the use of AI, according to December 2022 and 2023 surveys, respectively, by a National Association of Insurance Commissioners working group on big data and AI.

Yet AI won’t take over the entire insurance enterprise, no matter how smart it gets.

“There will be humans in the loop,” said Chris Raimondo, the Americas consulting insurance leader at EY.

Here are some changes you may see when dealing with your insurer and why AI advances won’t move as fast in the insurance industry as today’s buzz implies.

As companies get better at working with AI, buying a policy and communicating with your insurer may get easier.

“Quotes will take minutes instead of hours,” said Michelle Youshock, head of personal lines at World Insurance Associates.

Chatbots will get smarter, with the potential to provide relevant answers instantly 24/7 and the ability to patch in a human customer service representative when necessary. And the application process will become more seamless, insurance AI experts say.

For example, traditional online forms have structured, drop-down menus and standard questions. With AI, a digital agent could “capture the customer’s needs based on the way they describe it, and then automatically fill out the forms for them,” said Matt Bonakdarpour, president and chief technology officer at Root Insurance.

AI tools can already help verify data on a policy application up front to make sure there are no missing or incorrect details, said Jeremy Jawish, CEO and co-founder of Shift Technology, an AI technology provider for insurers. “That means fewer surprises later, like finding out your coverage doesn’t apply because of a paperwork error.”

Meanwhile, new AI tools are helping underwriters collect information and assess risk more quickly. For example, Flyreel, a tool for insurers from LexisNexis Risk Solutions, lets homeowners use a self-guided mobile inspection app to capture details inside and outside the home, including the roof, often eliminating the need for an insurance company to send out an inspector. AI evaluates the images and flags potential risks.

“By reducing the need for boots-on-the-ground, on-site inspections, AI tools are helping carriers improve risk assessment, streamline underwriting, and speed up claims,” said Brian Keller, vice president of data science and AI at LexisNexis Risk Solutions.

Learn more: Homeowners insurance: What it covers and how much you’ll pay

Insurance companies have been using machine learning, a type of AI, for years to sift through data and help make predictions and price policies.

Advances in generative AI, which can digest information and create content, enable insurers to transform even more data into a form the models can use, Bonakdarpour said. That helps insurers price policies more specifically, rather than basing prices across broad categories.

It could also make coverage more available in some cases. For example, landlords traditionally paid exorbitant prices or had trouble getting insurance for older structures because the age of the building put them in a high-risk category. AI tools help real estate insurers like Honeycomb Insurance evaluate individual properties more closely and offer coverage to properties that are well-maintained.

Without AI, a human underwriter could do that, but it would require spending a lot of time on individual properties, which isn’t cost-effective, said Roy Feig, Honeycomb’s senior vice president of product. “AI understands a ton of data points and can get to an accurate decision very fast.”

Insurers are also beginning to gear coverage to real-world data, such as driving behavior from telematics technology and smart-home sensors, Jawish said.

“In the next few years, that could mean insurance that adjusts in real time to policyholders’ behavior,” he said. “Safe driving, routine maintenance, or even installing a smart-home device might earn them better rates or prevent losses before they happen.”

Learn more: Car insurance discounts: 17 ways to save

For simple claims, like car damage after hitting a deer, a claim can be settled almost immediately using AI and information like photos, location, and telematics data from the car, Bonakdarpour said.

For complex claims, such as car crashes involving injuries and multiple drivers, insurers are using AI to streamline the process and shave days off the time to settlement. These claims naturally take longer because claims adjusters must review huge volumes of information to make decisions. A settlement demand packet from a personal injury attorney alone can run 50 pages or longer and contain loads of jargon.

“Now AI is summarizing it into a one- to two-page document that’s clear and precise, which is allowing adjusters to make decisions quicker,” said Chris Bakes, managing director of auto solutions at Sedgwick, a claims administration provider for insurance companies.

Meanwhile, adjusters can use AI virtual agents to gather all the data. Traditionally, human adjusters had to log into multiple different sources and apps, search for the information, and compile it themselves, Raimondo said.

Today, virtual AI agents are assistance tools, Raimondo said. “I think at some point, we’ll move towards them being actually executors and orchestrators of workflows on behalf of [human adjusters].”

Learn more: How to file a car insurance claim

A number of things will likely keep AI in check.

Insurance is highly regulated, with each state having its own laws and rules governing the industry.

“There’s a lot of requirements, and you have to make sure you’re staying compliant and aligned with the regulatory agencies,“ Raimondo said.

Privacy and bias concerns

AI depends on a ton of data from many sources, and insurance companies have to keep it secure.

They also have to make sure the algorithms don’t unintentionally introduce bias or make opaque decisions, Youshock said.

Insurers will need to be able to explain how consumers’ data is used, when automation is involved in claims, and the reasoning behind decisions, Jawish said.

AI is only as good as the data it gets. Even the best systems can get confused and make mistakes, Jawish said.

And the stakes are sky-high because the cost of making an error is huge for both insurers and consumers.

That makes testing, controls, and human involvement essential. AI can help improve efficiency, but humans must still be in charge of the final output and decisions, Keller said.

“The best insurers use AI to assist experts, not replace them,” said Jawish.

Tim Manni edited this article.