Three Ways AI Can Transform the Claims Experience

The claims experience is a critical moment of truth, and one bad experience can cause customers to switch carriers. The growing role of AI in the insurance sector can completely transform claims or service rep conversations with customers, driving efficiency and delightful customer and employee experiences through an insurance organization.

AI Can Improve Response Times

In a recent study of what drives customer satisfaction in claims, Hi Marley found that timeliness of service and resolution is one of the top drivers of a 5-star customer experience. However, 21 percent of 1-star surveys occurred when a policyholder had to wait too long for a response.

While adjusters handle multiple cases simultaneously, AI can help employees prioritize which customer to handle first based on interpreting the nature of the incoming request, such as whether a policyholder is unhappy, if a question has been left unanswered for too long, or how severe the claim is.

AI Can Influence the Arc of Conversations

Furthermore, Hi Marley’s study revealed that the adjuster’s attitude and approach have the most significant impact on customer satisfaction. AI can also assist in this area. In terms of the conversation itself, AI models can be trained to coach employees by offering suggested responses to customers or the next best action for the employee based on what has driven positive outcomes historically.

This can help inexperienced reps avoid errors and steer conversations in a way that delivers the best customer experience, ensuring customers feel heard and provided for throughout the claims process.

AI Can Create Efficiencies at FNOL

Another opportunity lies with First Notice of Loss (FNOL).

Hi Marley’s recent survey found that policyholders want to use their voice to tell their story, with nearly 70 percent of respondents initially reporting their incident via a phone call, either to their agent or carrier. With this channel, customers can easily share every detail about what happened to them —how the accident occurred, their property or automobile status, if they were injured, and more. The claim is typically then triaged (manually or via automation) to an adjuster, who must collect photos, build an estimate, identify fraud, and more.

Generative AI can automate all of this in a single step.

AI can interpret a customer’s incident story and structure its data, request and evaluate photos, and detect fraud by deploying sophisticated models, all allowing for immediate and accurate triage.

When used to automate processes like photo collection and evaluation, AI can reduce cycle times. Hi Marley’s survey found that 59 percent of claims were resolved within two weeks when respondents could send in photos and videos with FNOL, compared to 31 percent without this option.

Also, AI can help with adjuster assignment and getting the claim to the right person the first time, thereby reducing claim evaluation time and improving the policyholder experience. Hi Marley found that 55 percent of respondents were very satisfied when they only had to work with one person to resolve their claim; if they had to work with two or more people, that number dropped to 39 percent.

However, because insurance is so nuanced, and AI is still relatively new, it’s critical that these models be trained on insurance-specific data and conversations, among other considerations.

Considerations for the Future of AI in Insurance

While AI presents enormous opportunities for the insurance industry, the regulatory environment for AI is evolving quickly. Many U.S. state legislatures are evaluating bills regulating the commercial use of AI. It’s, therefore, incredibly important to stay on top of this legislation to ensure compliance.

Two specific ethical and regulatory challenges come to mind for the insurance industry:

• Data accuracy: First, providing customers with correct data is critical in the insurance industry, and AI isn’t currently 100 percent accurate. The potential for hallucinations—when AI systems generate incorrect information based on patterns in the data they were trained on—is well documented. Putting generative AI directly in front of customers, whether through chatbots, marketing or policy content, or next-best-action modeled recommendations, risks misleading the policyholder or providing a poor experience. Most carriers implement certain checks, such as a human in the loop or other verification checkpoints, to catch errors or inaccuracies before they reach a policyholder.
• Bias in AI model output: The existence of bias in the output of AI models is another concern in the insurance industry. Much of the pending AI legislation directly concerns this risk and calls for creating AI task forces to navigate this potential harm. For example, bias could inadvertently be applied in fraud modeling or liability determination in claims. While inputting biased data into AI models is a clear red flag, it’s not always apparent which data contains bias; it may be that the way that the AI model combines the data is what causes bias.

Insurance is still a people-oriented, people-focused business. Carriers often deal with policyholders during some of the most stressful events they’ve ever experienced. So, with all of the benefits of AI, it’s crucial not to lose the human element of interactions entirely.

Carriers that combine the benefits of AI and personal touch will have the most success in navigating this new technology and transformation.