Spotlight: Can GenAI fix insurance claims headaches?

Generative AI emerges as a promising solution for claims processing, enhancing efficiency and customer experience amid rising volumes and operational challenges. Chris Marshall reports.

More than two and a half years since ChatGPT brought generative AI (GenAI) into the mainstream, insurers are still grappling with how to make the most of its potential.

There is certainly a demand for technological solutions to challenges including rising claims volumes. Nearly two-thirds (65%) of insurance sector respondents to a 2024 Reuters survey stated that leveraging technology to increase operational efficiency was the best approach to addressing rising costs associated with claims.

GenAI, the type of artificial intelligence that can create new content and ideas, is at the forefront of this technological promise. Deloitte has described it as “potentially the most powerful tool available to UK insurers to help transform their customer experiences and outcomes”.

But adoption is not simple in an industry burdened with legacy technology and manual process, as well as acute legal, ethical, and regulatory scrutiny.

I used to be a claims handler, and I know how much correspondence and complex documents they face – the capacity of GenAI to summarise it means a huge gain of time.

Rémy Gounel, subject matter expert, claims at Shift Technology

Beyond the script

The opportunities for GenAI in claims processes and beyond are many and varied. In some cases, it is already starting to make a tangible impact.

One area considered ripe for improvement is claim handlers’ scripts. The same 10-question script for claims handlers managing a storm event may not be right for every situation.

“Rigid claims scripts and manual processes no longer fit the complexity of modern claims – whether it’s a storm-damaged property, a motor collision with layered liability issues, or a business interruption claim,” says David Scott, partner at HF and managing director of HighFive.

With GenAI, insurers can get personal. They can access multiple data sources to ask the best questions, not just what a script mandates. As well as saving time, it means customers are saved from the annoyance of being asked the same question multiple times throughout the claims journey.

Rory Yates, global strategy lead at EIS, says the automation of call scripts is among the best use cases he has seen for Gen AI, helping call-handlers quickly understand a customer’s context and respond in a more informed, personalised way.

But GenAI’s use cases go far beyond scripts. Rémy Gounel, subject matter expert, claims at Shift Technology, says GenAI can act as a tutor for junior claims handlers, offering expert guidance to improve decisions. It also supports experienced staff by summarising complex data, such as photos, emails, voice notes, helping optimise both time and the quality of responses to policyholders.

GenAI is also being used to translate languages in real time, to triage claims more accurately and suggest the most relevant claims pathway based on the likely size and complexity of the claim. It can help claims handlers with drafting correspondence and assisting in valuations.

GenAI’s ability to manage correspondence and digest complex documents is also transforming the day-to-day work of claims teams. Gounel says: “I used to be a claims handler, and I know how much correspondence and complex documents they face – the capacity of GenAI to summarise it means a huge gain of time.”

Consultancy Capgemini points to the use of Gen AI models in personal property claims that incorporate real-time information to update its recommendations. Phone call audio is converted to text and fed into the model.

Insurers are at different stages of adoption. Take the example of Zurich in the UK. The insurer is deploying GenAI applications to ingest and classify claims documents much quicker. These applications are helping to improve the handling of first-notice-of-loss data, and support more accurate classification of data at intake.

Zurich is also using AI to identify patterns in claims data, such as loss adjuster reports, which can enhance both claims and underwriting decision-making. By analysing large volumes of information, these tools help teams identify emerging trends, and make more informed, data-led decisions.     

Laying the groundwork

Zurich perhaps makes it sound easy. But first insurers must, as Yates puts it, reset their “archaic technology foundations”. 

Legacy technology isn’t the only challenge. Insurers experimenting with GenAI must grapple with considerations including use case prioritisation, risk assessment and regulatory compliance.

In response, there has been a surge in investment by insurers, according to Capgemini. Adam Denninger, global industry leader for insurance at the consultancy, expects to see major pieces of claims processes being reworked over the next one or two years. But it is still very much early days. “The simple reality is that most insurers are not using generative AI at scale or broadly across their core processes – in claims or elsewhere,” he says.

“Some have started various proof-of-concepts for specific use cases, a few have focused solutions for relatively limited scopes in production right now, but many have not started, and frankly, they often don’t know where to begin.”

Extensive up-front investments are needed, according to Denninger. For example, AI is only as good as the data it is working with – so insurers must invest in cleaning up their data before being able to derive large-scale value from AI. This must come alongside organisational change.

We do see GenAI as promising when it comes to extracting insight from unstructured data such as emails, voice notes, adjuster reports and legal correspondence.

Richard Napoli, claims and legal services director at Markel

For Helen Rogers, head of claims digital experience at Zurich, the wider challenge is identifying the right use-cases to focus on – “Those that make the most difference to both our business and customers first.”

Juggling this need for improving the customer journey, while maximising the efficiencies afforded by Gen AI is a fine balance to strike.

Yates likens this orchestration to air traffic control: a dynamic environment where data must inform decisions quickly and accurately, and where humans can always jump in when journeys go off track.

For business insurer Markel, the focus remains very much on the human touch. Richard Napoli, claims and legal services director, explains that most claims Markel receives are third party. “Our job is to protect our policyholder and guide them through the stressful situation. Having an empathetic and knowledgeable claims professional to talk through their situation is almost always welcomed,” he says.

There is still a place for GenAI for the firm. Napoli says: “We do see Gen AI as promising when it comes to extracting insight from unstructured data such as emails, voice notes, adjuster reports and legal correspondence.”

Another challenge for all industry players is the quality of data being fed into GenAI models. While insurers don’t lack for data, much of it is unstructured and turning it into insight can be challenging. “This is made worse, because most operational data models are centred on policies not customers,” says Yates. 

“As a result, applying various data points across claims, interactions, policy changes, mid-term adjustments, other digital experiences, and call logs, and so on becomes difficult to centre around the customer in real time without building old fashioned data-lake models.”

GenAI can have huge positive benefits for customer experiences, but it needs to be fed good customer data.

Such challenges are slowly being chipped away. Gen AI use in insurance seems a question not of ‘if’ but ‘when’ and ‘how’. Andy Hutchinson, chief revenue officer at Turvi says: “I believe we will get to a point in the future where it will be hard for us to remember not having artificial intelligence embedded into our claims processes.” 

Sponsored content