AI's impact on claims processes and procedures – Part 2

Part 1 of this two-part series examined how carriers are implementing AI into different aspects of the claims process such as first notice of loss, customer communications with policyholders, verifying coverage and synthesizing data associated with claims.

It's all about the data

Insurance claims encompass vast amounts of data, and AI's ability to assist with summarizing information can help streamline the process for adjusters and customers.

"The real power of AI lies in how it synthesizes data to support smarter, faster decisions," affirms Peter Miller, CPCU, president and CEO of The Institutes. "AI can pull in data from multiple sources—photos, weather feeds, repair shops, vehicle diagnostics—to generate and verify estimates in real time. It can automate verification, checking claim details against coverage language, previous claims, and even external records. It flags high-risk cases and automates routine ones—so adjusters can focus their energy where human judgment and critical thinking adds the most value."

AXA XL's Pam Urueta, general counsel and chief claims officer, concurs. "We've seen value in using Gen AI to help summarize documents, extract data, and provide initial drafts of communications, freeing them up to focus on higher-value work, like engaging with clients and using their expert judgment." 

The data derived from claims can also enable insurers to be more proactive in helping customers prevent recurring losses by using technologies like Internet of Things (IoT) sensors, identify and mitigate existing risks and highlight properties or issues that may require specific endorsements. It changes the role of insurers from being simply reactive to being proactive, saving time, money and resources for everyone. 

The increased use of AI in claims presents several considerations and opportunities. Fraud continues to escalate and while AI tools help in its detection, they also embolden bad actors who use AI to enhance their efforts. 

Data quality is an issue because some of it is inaccessible or inaccurate, which affects the final outcome AI generates. The actual systems carriers use present another issue.

"The biggest challenge we're seeing is integration complexity," explains Near Space Lab's co-founder and CEO, Rema Matevosyan. "Many insurers have decades-old infrastructure that wasn't designed to handle the volume and velocity of data that AI systems can process. There's also a significant training curve for the human workforce to understand how to work alongside these systems effectively."

Matevosyan continues, "From our aerial imagery perspective, we're also seeing challenges around data standardization. Different AI systems expect data in different formats, and there's still work to be done on creating industry-wide standards."

AI adoption comes with risks

For all of its promise, the adoption of AI into the claims space could create some risks for carriers and other companies utilizing this technology. Government regulations (state and federal) and how the legal industry addresses AI-related lawsuits will also impact companies' integration of AI.

"The main challenges are the use of AI and the litigation risks associated. This needs careful consideration with a human in the loop process and controls, who can manage any risk from the deployment of AI, and can ensure that we protect ourselves and our customers," shares Aspen's chief operations officer for claims & UK operations, Lee  Elliston. "The other main challenge is change management. To truly get the benefit from AI, it requires skill, practice and process change. This is not solely about deploying technology and AI quickly, because the AI insurance train has left the station."

The Institutes' Miller expresses concern about transparency and explainability, which he believes "are critical—especially for adjusters and regulators. If AI is going to support or make decisions, the logic behind those decisions must be transparent."

How policyholders view the integration of AI into customer-facing processes is also a consideration. Multiple studies conducted by Digital Insurance have found that customer experience, satisfaction and retention are among the key drivers of technology innovations for insurers. Some uses will be highly visible such as chatbots to answer questions or personalized claims-related messages to update customers, while others will be more invisible because they are behind-the-scenes integrations.

"In commercial lines, most insureds may not be directly aware that AI supports claims processing, but we see it contributing to a better customer experience: faster processing, faster communication and clearer documentation," says Urueta. 

Miller concurs. "Policyholders care most about clarity, speed, and fairness—and when AI, with human oversight, delivers on those, the feedback is positive." He cites the ease of uploading photos, getting real-time updates and being paid within days as examples. "There is still some wariness around 'robot decisions,' especially for denied claims. That's where human, plus machine models—AI-assisted but adjuster-approved—are striking the right balance."

Matevosyan is encouraged by what she's seeing in terms of how policyholders are reacting to the integration of AI into the claims process. "While some people are always skeptical of new technology, what we find with our customers is that as policyholders experience these faster outcomes, they become increasingly supportive of insurers using AI…What's particularly encouraging is that AI is making underwriting and claims more transparent than ever before."

Claims is just one of the areas where carriers are integrating AI into their processes. It takes time, training, support from employees and customers, and transparency to steer a successful implementation, but the outcomes will far outweigh the risks.

"Ultimately, AI, with proper human oversight, is helping insurers shift from being just a backstop to being an active risk partner—one that can predict, prevent and fulfill the promise of insurance," concludes Miller.