Newsletters
Risk Central
Digital EditionHow AI Is Changing the Claims Landscape: Speed, Accuracy and a Human Touch
AI technologies are advancing rapidly and changing many aspects of the insurance industry, particularly in claims management. Generative AI applications can improve efficiency, communications, and customer experience by getting to resolutions faster.
However, adopting AI is not without challenges.
Experts note that AI outputs are only as good as the data they work off of and that companies must continually train and refine their models for the best outcomes. Insurers also need to know the limits of their AI, designate leaders to manage and govern the technology, and ensure that it is used with the right balance of humans in the process.
What is clear is that AI is moving so rapidly that stakeholders who are not keeping up will fall behind fast, said Jeff Gurtcheff, chief claims officer at CorVel Corporation.
“In some respects, there is a race to see who’s first to use it to its full potential. It will enable adjusters to handle a lot more than they have in the past. Companies that do will have significant advantages not just in results but in profitability and efficiency,” he said.
Big Opportunities in Claims Management and Handling
AI has experienced tremendous growth in recent years and is projected to contribute nearly $20 trillion to the global economy by 2030, according to IDC.
Generative AI, a subset of AI that goes beyond analyzing information to creating content like text, images, and video, has opened AI to the masses with more applications. Software developers now use generative AI chatbots like ChatGPT and Microsoft Copilot to provide AI applications to nearly every industry, from banking and real estate to health care and hospitality.
While insurers are piloting and deploying AI applications in several areas of the industry, one promising space has been claims management. Analysts at Bain believe generative AI offers a $100 billion opportunity in P&C claims handling by reducing loss-adjusting expenses by 20-25% and leakage by 30-50%. To make the most of it, insurers will need to identify where to best focus their efforts and address organizational changes that will be required.
Kenneth Tolson, CEO at Turvi, notes many insurers are now using some level of AI in “nearly every aspect of the claims ecosystem,” from First Notice of Loss (FNOL) through adjudication and the auditing of payments. Many also use AI to guide conversations, helping them ask better questions and summarize the interactions. From initial notices to keeping in touch with customers and processing the payment, AI ultimately helps claims departments get to settlements more quickly.
“It ultimately drives a better customer experience, whether that’s through the speed of accelerating the process or actually interacting with customers themselves,” said Tolson.
Improving Efficiency, Communications and Decision Making
The most immediate impact AI applications have in claims management is efficiency and taking the manual and time-consuming work out of organizing notes, documents, calls, and information, said Joe Powell, chief digital officer at Gallagher Bassett.
AI chatbots like ChatGPT and Copilot can summarize large amounts of information in seconds, helping adjusters quickly get up to speed on a particular aspect of a claim.
“It makes it so much easier to make gold out of all that information and find the important nuggets buried in that pile of unstructured data. AI can significantly speed up the process,” said Powell.
AI applications can also listen to calls to generate transcripts and create organized summaries for adjusters and stakeholders. For example, AI-based phone solutions like Regal.ai and Talkie.ai enable customers to initiate claims through voicebots, which can field the right questions, provide answers, and relay information to human adjusters.
Powell noted Gallagher Bassett is also developing technology for inboxes that identify critical emails and flag them for handlers to ensure they’re addressed.
Applications like these can yield significant savings in the claims process by reducing the time spent on manual processes and administrative tasks, he said.
However, Gen AI can also be used in more advanced applications. For example, machine learning models can help make important claim decisions, such as identifying the reserve on a claim or if a workers’ compensation claim requires a nurse, case manager, or anyone with clinical guidance on the clause. AI models can also analyze trends and consider data attributes to flag potential cases of fraud, said Powell.
“All of these things are areas where a decision could benefit from a second set of eyes. The AI models can flag cases and make recommendations to help the adjusters,” he said.
Some insurers already use AI in straight-through processing for high-frequency, low-severity claims. These AI agents can streamline processing with automated document handling, real-time decision making, and personalized communications with policyholders.
Joe Powell, chief digital officer, Gallagher Bassett
This frees up humans to focus on more complex claims.
Turvi’s CoverAI application uses AI, analytics, and advanced risk insights to help adjusters automate the claim coverage review process by quickly interpreting policy wordings to accelerate decision-making. It also helps adjusters more consistently interpret policies, improving accuracy and reducing human errors.
Information captured in the systems also enables insurers to gain more insight into coverage issues and offer additional guidance for underwriting. Other AI claims adjuster tools on the market include Clive and EvenUp.
Gurtcheff notes the next level of AI is discriminative AI, which classifies data and supports more advanced models to identify patterns and predict outcomes. Corvel began using the technology to support prescriptive models that can recommend responses to anticipated scenarios. For example, it may identify a claim that is more likely to be litigated, state the reason why, and then propose several solutions.
“A prescriptive model doesn’t just tell you about what happened, but what you can do about it. It prescribes certain actions the adjuster can take,” Gurtcheff said.
Mitigating Risk with Careful Implementation and Model Design
Despite its benefits, AI carries risks and challenges.
Insurers have to implement applications in a way that protects intellectual property and consumer data and is compliant with all industry PPI standards, said Tolson.
Insurers must also understand AI’s limitations, ensure data integrity and check the conclusions of algorithms. One particular risk is hallucinations, a phenomenon where AI gives an incorrect or misleading response, often due to an incorrect prompt or improper training.
“It comes down to implementation and whether you are training the model appropriately. You have to be thoughtful. If you move too quickly with AI, you create more risk,” said Tolson.
AI demo applications must be highly accurate and tested before they are put into production, said Gurtcheff. One best practice is to test scenarios with AI models against human experts to see how they compare in terms of quality and accuracy.
AI is not a “set it and forget it” technology but an iterative process that requires constantly training and retaining models to ensure accuracy and prevent hallucinations, said Gurtcheff.
Change management is also essential, and insurers should consider establishing roles to manage AI governance and training employees on using new technologies. It’s best to have a specific lead designed to manage the AI, said Gurtcheff.
“Within our data sciences group we have people that are specific to those functions of training and retaining the AI,” he added.
Finally, insurers must learn to strike the best balance between AI and human associates, recognizing both strengths and weaknesses. Gurtcheff noted that in the future, it is likely that most low-severity, high-frequency claims will be fully automated.
However, as insurers still rely on relationships and empathy, humans will always play a critical role in claims resolution.
“If you’re an adjuster, part of that is demonstrating compassion, empathy, and effectively communicating with people. There’s that balance of using AI and automation but also having human empathy and experience,” he said. &
