(Datos Insights’ recent Insurance CIO Council event, hosted concurrently with the InsurTech Hartford Symposium. Source: Datos Insights.)
At Datos Insights’ (formerly Aite-Novarica Group’s) recent Insurance CIO Council event, hosted concurrently with the InsurTech Hartford Symposium, over 70 insurance executives spent two days with us talking about the technology topics that are top of mind for them. Unsurprisingly, artificial intelligence (AI)—and ChatGPT specifically—was one of the themes most discussed. This has proven true across the other industry events we participate in as thought leaders, such as those hosted by LIMRA and AWS.
Realizing that your organization must come to terms with the new AI-enabled world feels like looking up at a mountain from its base. The first stop for insurers on this journey is near the base of the mountain: They can see the activity and storms above, but they haven’t started to climb yet.
The second stop is a little further along, where cloud cover causes confusion and doubt. At this point, there is a strong sense of needing to head somewhere, but no certainty as to what direction to go. There are “experts” on every side making suggestions, but they’re as likely to hinder as to help when you’ve lost track of the way forward.
The third stop is where insurers start to ideate. This ideation tends to be very broad; while it may not lead directly to action, it often leads to more focused ideation. ChatGPT and similar tools are constantly front of mind, perhaps prompted by interest or questions from the CEO or the board. At this stage, there is a lot of activity, but little to show for it—despite being a necessary part of the journey.
The fourth—and, at this point, final—stop involves active experimentation and implementation. This far up the mountain, the views may be intoxicating, but the risks have also risen. Here, insurers are actively engaging with proofs of concept while they try to determine the cost and value of a full-scale implementation. Very few projects have moved fully into production, so the true pinnacle of the journey has yet to be reached.
In our experience, the population of these four stops varies greatly. Client engagement leads us to conclude that the ideation stage is by far the largest, followed by active experimentation. The number of companies at the base of the mountain, watching the confusion above but not yet moving, appears to be quite low; although this is an unusual situation for insurers, these are unusual times.
Considerations for Your AI Journey
ChatGPT does not equate to AI. AI is a multi-dimensional and deeply complex topic. There is no single end point common to all insurers; it will depend on each insurer’s needs, appetite, and risk tolerance. From our conversations with carrier CIOs, insurers for the most part understand this. Insurer sophistication on this topic is increasing quickly.
Yet some boards and executive leaders may continue to conflate AI and ChatGPT. Solution providers may not be providing sufficient clarity on how transformer technology—the “T” in ChatGPT—fits into the larger mosaic of AI. Transformer type technologies are wonderful considerations for the proper problem, but also totally inappropriate for some use cases we’ve heard others suggest.
There are serious considerations that permeate the entire landscape of AI. Considerations such as cost, security, an unclear regulatory environment, and impact on the employee base should be given serious thought. Clarity on these topics may emerge over time and add a significant risk, depending on the choices made.
Navigating the World of AI
There are innumerable emerging best practices and learnings that can be leveraged by carriers as they begin or continue their journeys. However, few of those can be broadly applied without first understanding each journey’s specific case. Here, we suggest two that are broadly applicable no matter where you currently find yourself.
First, bring a guide or two on the climb. No insurer should think they have to go it alone. The domains of AI, not just ChatGPT, have been invested in for years, even though we’re just starting to see movement. This is a field that can change rapidly; by year end, it’s possible we won’t even think of 2023 as the year of ChatGPT, but some other yet-to-be-released solution that’s far more impactful. Having a guide to help assess and climb safely will be invaluable in turbulent conditions.
In addition, to further avoid “going it alone”, insurers should look to their solution providers to understand how they are investing in AI. Insurers should not be attempting to build all AI solutions themselves. Insurers need to be conversant enough in the AI domain to evaluate vendor readiness and if the investments are expected to accrue value to the insurer, which again reinforces the value of a guide.
Don’t go it alone. Keep your eyes on the horizon for incoming storms.
For a more in-depth discussion of the impacts that ChatGPT and other generative AI solutions will have, read our recent report “ChatGPT in the Insurance World: Chat of Things to Come” where we share some initial perspectives and insights on how insurers can explore, monitor, strategize, and modernize with ChatGPT in mind.
Unlocking the Future of P&C Insurance with the Power of ChatGPT
ChatGPT and the Future of Insurance with Large Natural Language Models