On April 19th, I jointly hosted the Aite-Novarica Women’s Network meeting with Head of Property/Casualty Insurance Martina Conlon on the impact of artificial intelligence (AI) in insurance. We were joined by Gwen Adler (Financial Services and Insurance Lead, Dataiku), Meredith Barnes-Cook (VP, Ushur), Aarti Gupta (Insurance GTM Head, AWS), Kim Wiswell (Senior Product Manager, GradientAI), and Stephanie Dalwin (Advisor, Aite-Novarica Group), who shared their perspectives on the use cases, challenges, and impact of AI in the workplace. Here are the key takeaways from our discussion:
AI Can Impact Nearly Every Aspect of Insurance
There are a significant number of use cases for AI in insurance. AI is being used to increase efficiency and effectiveness in all areas, including product and actuarial, marketing, underwriting, customer service, billing, claims, and compliance. When finding ways to incorporate AI into your organization, it’s important to look for use cases that can quickly deliver business value and build from there. AI is best used to augment insurance tasks and should not be aimed at replacing human expertise.
One panelist identified underwriting as a key area for automation. During the proposal stage, speed matters. Finding ways to automate the entry of submission data can not only alleviate workloads, but also improve risk selection as companies are now able to better leverage all of the submission data rather than just the data entered. Another example our panelists brought up was in claims. On the claims side, and in workers’ comp specifically, insurers are utilizing texting capabilities early on in the claims process to gather vital information while keeping the tone conversational, and often empathetic, enabling them to resolve claim issues more quickly.
Make Sure the Right People Are Involved in AI Projects
When beginning a new AI project, it’s important to determine the expected ROI and how you plan on sustaining that value long term. Understanding how to get a project out of the idea phase and into execution mode is an important skill set for those involved with AI. Additionally, it’s critical to ensure that SMEs are included in the conversation. SMEs work with the data every day and can provide the proper context and business needs to enable IT to build solutions that make tangible impact.
One panelist discussed how organizations can build frameworks and models that are scalable and applicable to the majority of use cases to prevent an organization from having countless, bespoke processes. Proper project prioritization is a key step. There are often competing priorities, and it’s difficult to know what projects will produce the most value, so focus efforts on those that have a measurable ROI and address a key business opportunity or challenge. This panelist also noted how important it is to bring SMEs into conversations to answer the business questions and provide key insights. Data people don’t always understand the context of the data they’re looking at, so bringing in experts to provide context can make the impact much greater.
ChatGPT Can Help With Tasks, but Expertise Is Still Needed
ChatGPT is a hot topic, and many organizations are looking to determine how it can be used to meet their needs. Current tests of the platform have shown that it has the potential to help with summarizing data and can provide significant time savings for laborious data entry tasks. However, it is critical to have someone with expertise check the output’s accuracy or an organization can risk running into a compliance issue.
One panelist mentioned summary data by nurses as a good use of ChatGPT. She mentioned that currently, nurses spend about four hours summarizing data, but ChatGPT can do the same summary in seconds. Another panelist agreed and mentioned that if you are prescriptive with what you give it, you can get a fast response, but that it’s not always accurate. In an industry where compliance is key, insurers need to find a balance between finding efficiencies and opportunities for automation while maintaining accuracy and compliance. While ChatGPT, and AI in general, can provide great opportunities for process improvements and efficiency gains, balancing technology with the human element remains imperative.
Our next Aite-Novarica Women’s Network virtual meeting, titled “You’ve Come a Long Way…Maybe,” will take place on June 28th at 2pm ET. We will be discussing some stats about the progress women have made in the industry broadly. This meeting will be open to both Women’s Network members and non-members. Register to join us here.