AI set to exponentially improve insurance efficiency

CHICAGO – Artificial intelligence technology such as ChatGPT will quickly lead to radical improvements in insurance underwriting and claims handling, insurance industry executives said.

Using AI to gather and analyze information will improve risk mitigation and allow insurers to take a more sophisticated approach to pricing, they said.

And the benefits will be so big that any efforts to slow the adoption of AI are likely to fail.

Large language models such as ChatGPT are not being used to change the business model for the insurance sector, but they will make current processes much more efficient, said Ericson Chan, Zurich-based group chief information and digital officer at Zurich Insurance Co. Ltd.

He was speaking at the Insurance AI and Innovative Tech USA 2023 conference sponsored by Reuters News & Media Ltd. in Chicago last week.

The insurance industry is already processing large amounts of information from claims documents, risk engineering reports and other documents, he said.

“In order to extract the information from tens, hundreds or thousands of documents, these models are very efficient,” Mr. Chan said.

The models, for example, can structure information on specific buildings, specific risks and other factors from many different documents, he said, adding, “It will save tremendous amounts of time.”

The models can also be used to update computer coding in the industry. For example, rather than educating a developer in a piece of code that was used 20 years ago to enable him or her to update it to modern code, the models can very quickly translate the legacy code into a modern programming language, such as python, Mr. Chan said.

Programming projects that would typically take a month can take half a day using large language models, he said. So-called 10x developers, who take one-tenth of the time that average developers to complete a task, are highly sought after in programming sector, he said.

“If we can do that with developers, what about claims adjusters? Can we have a 10x adjuster? Can we have a 10x underwriter?” he said. “It’s not changing the business model, but it makes our current insurance business a lot more efficient.”

AI will also be used to eliminate weaknesses in the underwriting process, said Dan Abrahamsen, CEO of Cover Whale Insurance Solutions Inc., a New York-based insurtech managing general agent specializing in commercial trucking coverage.

Cover Whale uses telemetry, or remote data collection, to gather driver behavior data and vehicle performance data from trucks it insures.

“We’re seeing results simply by engaging with our customer and identifying and trying to prevent the high-risk driving behavior,” Mr. Abrahamsen said.

The insurer is also able to use the data, in conjunction with traditional underwriting, to offer discounts to lower-risk drivers, he said.

By using AI to analyze data provided through technology such as telemetry, insurers will be able to reduce human bias in underwriting, Mr. Abrahamsen said.

While the technology will replace some jobs in the insurance sector, its biggest effect will be on how it changes the jobs of many workers, he said.

“The scariness to me is it’s moving much quicker than I think most people are anticipating and that’s only going to continue. I think humans are pretty bad at understanding exponential effects as a group,” he said.

But efforts to slow the adoption of AI, such as the recent open letter signed by several technology CEOs asking for a six-month halt in AI research to establish safety protocols for the use of the technology, are unlikely to be successful, Mr. Abrahamsen said.

“I think it’s more of an arms race, and it’s going to be treated that way,” he said.