2024 has been the year of AI. The emergence of generative AI capabilities (GenAI) at scale last year caught insurers’ attention in a big way, and a huge portion of industry discussion since then has centered on AI: understanding what it does, speculating on how insurers might apply it, and (sometimes) piloting programs to see if those ideas prove out.
As we near the halfway mark of 2024, most insurers feel they’re behind in their preparedness for GenAI. But they’re very, very eager to experiment.
A study conducted earlier this year by Datos Insights advisors Mitch Wein and Ashley Lee found that most insurers are early in their AI journeys. Many have taken few steps (some haven’t taken any steps at all), and almost none have turned attention to more formal aspects like governance.
Some insurers are already experimenting with GenAI, though, and looking to do more. We found in our annual Budgets and Projects studies that more than a third of P/C insurers and about a quarter of LAB insurers plan to pilot LLMs in 2024. That’s in addition to the ~30% of P/C insurers and ~15% of LAB insurers who already have an LLM in-house in some form.
At least for right now, “using an LLM” generally means the insurer has set up access to an enterprise version of a commercial LLM (e.g., CoPilot, ChatGPT, Claude) and set out some ground rules for using these tools in a way that doesn’t expose company data. Business users can then play with these LLMs to see how they might optimize workflows and make tasks easier.
What will the next steps for insurers be? There will be a few paths:
- Using a commercial LLM: Some insurers’ first step, particularly for those who haven’t yet piloted an AI, may be using a commercial LLM as described above. Users will be able to find creative applications and ways to save time.
- Customize an LLM: Larger insurers may license an LLM instance or use an open-source model, then train it on internal data, perhaps for tasks like document review or natural-language forms search.
- Get capabilities from a solution provider: Technology leaders and insurtechs who sell products to insurers are already working to incorporate GenAI capabilities into their solutions. This is likely to be a fruitful route for smaller insurers who can’t dedicate resources to internal AI groups.
No matter what approach any particular insurer takes, 2024 is going to be a year for learning. GenAI is a technology with a great deal of potential, but no one yet knows what its most effective applications will be.
Insurers who haven’t started using GenAI yet will be happy to know many of their peers are also early-days, but they should start planning pilots now so as not to fall behind the curve. Experience gained in smaller pilots will help shape more specific applications that have the potential to have greater impact on efficiency.
Datos Insights Senior Principal Tim Baum and I recently published a brief on the state of play in insurer plans to use GenAI in 2024. For more information, download the report or contact me for a further conversation.
