Preparing Insurance Carriers and Brokers for Generative AI

In our last blog on AI innovation in insurance, we had a little fun with ChatGPT by asking it about technology trends in the industry. In this piece, we’ll focus on applying generative AI to the insurance industry today rather than using it to forecast the future.

Much of today’s discussion around AI starts with personal productivity and quickly moves toward the implications for an organization when individuals leverage it. As I always recommend to folks new to the world of generative AI, start by experimenting with something you know well and use it to get an accurate understanding of the limits and capabilities of the technology.

In our recent webinar, The ChatGPT & AI Revolution: An Executive’s Guide, we cover additional exercises and use cases for integrating ChatGPT as a productivity solution at the individual and organizational levels.

With the widespread use of AI technologies exploding, most insurance businesses are currently in a stage where they’re self-assessing if they are ready to adopt these capabilities and determining how they can provide them to their workforce responsibly. But once you’re ready to safely apply the technology, where do you start? We’ll cover some ways insurance carriers and brokers can apply AI to their existing processes, as well as criteria for evaluating which projects are the most viable for this.

Focusing on the Now Versus the New

Once you’re prepared to implement AI at an enterprise level, you need to understand where and how you can and should apply it. Traditionally, many emerging technologies and innovations have been associated with full solutions that solve business problems in their entirety. This is particularly true in insurance, where InsurTech innovations are typically full solutions targeting a specific part of the value chain.

But with AI, the best use cases come as an integrated part of comprehensive solutions rather than a replacement of them – focusing on “the now” (improving current projects) instead of “the new” (a top-to-bottom process overhaul). Where can you supplement these tools to help with problems you’re already trying to solve? Consider the following criteria to help evaluate which of your current initiatives are the best candidates for AI optimization.

Organize Information

Is there a component of the solution that involves structuring information in a consumable way? Often, the work involved in collecting and organizing information into consumable outputs can be very manual and time-consuming. Generative AI can aid in this process by analyzing a large volume of text data and grouping similar content into clusters based on topics or themes, categorizing the information for easier access and retrieval.

For example, suppose you’re already focusing on reducing the pain of inconsistencies and inaccuracies in submissions from brokers. Why not use generative AI as part of that solution to take disparate pieces of information and structure it more consistently and more readably for underwriters?

Digest Large Amounts of Data

What processes involve individuals analyzing and working with large amounts of data? If you need to pour through repositories of documents or spreadsheets, generative AI can help summarize or find important information faster and more accurately than a person can. It can extract specific things like names, dates, locations and numerical data from unstructured text to streamline data collection and analysis.

For example, if you’re trying to streamline your claims authority approval process, you could use a large-language model to summarize the claim that a manager can easily consume to quickly determine if it warrants payment approval.

Ideate and Research Interactively

Finally, these technologies can present information more interactive and iteratively, boosting user experience and efficiency. AI-driven virtual assistants and chatbots can guide users through complex information, providing interactive explanations and answers to user queries.

For example, if you’re struggling with leveraging knowledge base articles for your core claims, policy or billing solution because it is too time consuming to find the right one, ChatGPT or a similar solution can provide an interactive and streamlined approach for obtaining useful problem-solving information in a more satisfying way across your workforce.

Conclusion

While you may currently be engaging in initiatives like the examples above, you are more likely to have a different set of ongoing projects. The key is to evaluate those existing efforts to similarly identify ways that a large language model can enhance the solution you are building. Eventually, you may consider more extensive and dedicated solutions, but until then, you are better off applying the technology as a supplement to existing in-flight solutions.

In the meantime, a simple exercise to consider the advantages of a large language model to enhance solutions you are currently contemplating can be a great way to introduce the technology into your ecosystem in an immediately beneficial way.