Insurers Are Adopting Gen AI like No Technology Before It

Insurers are always actively monitoring new technologies but are typically cautious in adopting them. Most carriers take a “wait and see” approach, holding off on investment until earlier adopters prove out a particular emerging technology, then greenlighting their own pilots or proofs-of-concept once it’s clear that new tech has value.

Generative AI isn’t working like that. Instead, GenAI is standing out both for how interested insurers are in it, and how quickly they’ve moved to adopt it.

By way of comparison, consider an established technology like cloud. Insurers were skeptical of cloud for a number of years before beginning to realize its benefits in the late 2010s. It’s only in the last few years that cloud has become the clear consensus for deploying new technologies, and most insurers still only have part of their infrastructure on cloud platforms.

In GenAI’s case, Celent’s research shows that 24% of insurers worldwide already have a generative AI application in production; based on pilot and proof-of-concept activity, we estimate that by 2026 that could jump to half. That’s about a quarter of insurers using a technology that’s been widely available for less than two years, with the potential for up to 50% penetration in less than four. By insurance industry timelines, that’s meteoric.

Here’s a second important way GenAI is distinct from other emerging technologies in insurance: GenAI is improving on itself very quickly, and what can be done with it is far more sophisticated than what could be done even a year ago.

With most technologies, a “wait and see” approach isn’t usually too detrimental. Waiting a year to migrate to a data lake or roll out a chatbot is generally not determinative. Insurance moves slowly.

GenAI may be breaking this rule too; the complexity of GenAI-based solutions is ramping up very, very quickly.

A year ago, insurers were ahead of the game if they had set up an internal sandbox for business analysts or developers to play with commercially available tools like ChatGPT. But just last week at Celent’s GenAI Symposium in New York, we saw multiple solution providers and insurers present and speak to in-deployment applications including RAG-powered appetite indication, litigation review, and a group benefit claims tool that hones notes and writes a first draft of a response letter to a claimant. In other words, the leading edge is well past using Copilot to help draft communications for agents or policyholders.

The next steps for insurers and solution providers will be to innovate on these uses of generative AI to solve even more complex problems. Further on, we’ll be seeing solutions that orchestrate multiple bots or multiple forms of AI to automate complicated processes. What GenAI can do is rapidly expanding in scope.

Experience is going to be a valuable building block as these more complex tools become reality, so for those insurers who aren’t yet engaging with Gen AI, it’s a good time to be experimenting. Please feel free to contact me if you’d like to start that conversation.