Life insurance companies have a reputation for being something akin to an oil tanker – big, and slow. But their approach to incorporating generative AI (GenAI) into various functions of their businesses can hardly be claimed to be thus.
A snap poll conducted by technology research and consulting firm, Celent, in June this year revealed that nearly 10% of life insurers in the United States were already using GenAI in production. And 50% said that they would be either using GenAI in a live environment or in a test environment before the year is out.
The technology is being used mainly in customer experience, sales and marketing functions, for a few reasons. Chatbots, for example, are arguably already embedded in these front office functions in many industries, not only insurance, so it’s not unusual for consumers to interact with GenAI through this medium.
But the tech is being used less so in middle and back-office functions, such as underwriting. According to Keith Raymond, Principal Analyst, Insurance at Celent, that’s because of the uncertainty around legal risk.
“Underwriting is a risk area for insurance companies when using AI,” he said. “Large language models are trained with data that can be biased. Biases in underwriting rules are a key point for mitigation, all humans are biased, consciously or not.”
The pace of change in the AI space has always been significant, but the release of GenAI provider ChatGPT in late November 2022 arguably thrust the technology into the public consciousness almost overnight.
Government responses vary widely across the globe although most are looking for interoperability they have differing opinions on how to balance risk verse innovation. In the United States, the Biden-Harris administration convened leaders of seven major AI providers in July to secure voluntary commitments from them around safety and transparency, and Vice President Harris convened civil rights leaders earlier in July to discuss risks posed by the technology in terms of discrimination.
These meetings were not insurance specific. But life insurers in the United States do have a yardstick by which to measure compliance when it comes to their use of AI. Back in 2020, the National Association of Insurance Commissioners (NAIC) published a document setting out principles which insurance companies should consider as they look to incorporate the technology into their businesses. But more recently, in July 2023, the NAIC published a Model Bulletin which specifically focuses on the use of large language models.
“The NAIC model bulletin is a big leap forward in terms of structure and guidelines for life insurers and their use of AI,” said Raymond.
These guidelines have been helpful, but best practice isn’t the law. And if Celent’s snap poll results turn out to be true – that half of all life insurers will have GenAI either working in a live environment at year end – then the customer experience, sales and marketing functions will still be the areas that see most of the investment, not just because of the efficiency gains this technology provides, but because of the lower regulatory risk.
And the middle and back office is arguably where the most gains are to be made from incorporating GenAI into a life insurer’s technology stack. While the front-end use might help a life insurer to shift a few more policies or save money on headcount due to the ability of the GenAI system to do a good enough job to replace a sales representative, the significant benefit lies with the underwriting function, and the ability for GenAI to mitigate mortality risk.
“Mortality risk is obviously a significant risk that life insurers are exposed to,” said Raymond. “Using GenAI in the underwriting process means that insurers will be able to manage mortality risk more effectively because they’ll be able to price policies more accurately. It’s an area of significant impact for an insurer.”
The full potential of GenAI is unknown, but what is known is that this is more a freight train than an oil tanker, hurtling onwards at high speed.
“The previous decade in the insurance technology space was all about the cloud – basically, migrating systems for efficiency reasons,” said Raymond. “But the next decade will be all about AI. For life insurance companies and regulators, this is a seismic change and it’s not slowing down. It could, in some ways, end up being the ‘forever decade’ for life insurers because of the enormity of the opportunity and the impact of it on not only the life insurance industry, but other industries that depend on it.”