AI is like a ‘co-pilot’ in the claims environment

The introduction of machine learning and the broader AI ecosystem into the claims environment is in its relative infancy—however, the potential it offers is immense. That is the message from Wayne Brand, head of Corporate Client Practice, Crawford & Company, who was speaking to FERMA Forum Today ahead of the conference in Madrid.

He noted that AI-based technologies are already supporting claims management on several fronts, from enhancing the customer experience and reducing processing times, through to improving efficiency and enabling richer data analysis.

“In many ways, AI’s status within the hierarchy is that of a co-pilot, supporting today’s increasingly connected adjuster in carrying out their daily tasks,” he said. “The technology can help alleviate many of the more resource-intensive tasks, automate some of the data-processing functions, and speed up the identification of errors or anomalies.”

He offered the example of a claims handler who makes a decision following a loss on an existing coverage or a policy endorsement. In its co-pilot role, AI can instantly review the decision of the adjuster by analysing the policy text against the scope of the claim to ensure that there is no potential conflict. “This action helps improve decision accuracy and reduce errors,” he said.

He offered another context: that of fraud. In this case, AI’s co-pilot role is that of identifying potential instances of fraudulent claims activity. “In current inflationary-burdened times, claims fraud has increased; yet, spotting it demands significant manual effort which can in turn stall processing times. AI can be trained to analyse claims data for possible cases with adjusters then only having to focus on those which have been red flagged,” he said.

These are two examples of a myriad ways that AI can be applied within the claims process to enhance each phase of the lifecycle, Brand noted. “Yet, if we are to realise its full potential then we need to reassess the data environment which feeds it.”

Data protection paramount

“AI cannot be allowed to operate without data borders.”

AI is at its most effective when it is operating in a data-rich environment, he said. Currently, while the claims management industry has more loss data at its disposal than ever before, access to data is strictly limited—and rightly so. Data privacy laws are increasingly stringent for a reason and adherence to these is a business imperative, he said.

“AI feeds on information—its ability to improve is commensurate with the amount of relevant data at its disposal. In the case of fraud, access to broader datasets will inevitably boost detection rates, reduce processing time and cost, and ultimately improve the claims experience for all,” he argued.

“But AI cannot be allowed to operate without data borders. There are more and more instances of how such technologies breach data protection systems in search of relevant information.”

He said that customers want clarity on how their data is being hosted securely in a more AI-exposed world and it is essential that all necessary measures are in place to protect it.

“So, when we have these two forces at play—the need for greater data protection and the need for greater access to data to help optimise AI capabilities—what can be done to achieve an effective middle ground?”

He notes that insurers already share some information with other insurers to prevent fraud and crime. “But perhaps there is more we can do to enhance AI capabilities. Every organisation within the sector maintains its own databases—given the size and scale of Crawford, for example, we have access to considerable claims information.

“Perhaps we need to explore in a more considered way how non-client-related information and insights can be made accessible in ways that can help further enrich the data that is available to the insurance sector as a whole.”

Underpinning this, there needs to be greater connectivity and networking of data, he believes. “We need to work towards an ecosystem which is as close to real-time as possible and from which we can augment our AI capabilities through improved access to relevant data in a controlled way.

“Clearly a debilitating factor given the global nature of the sector is the divergence in approaches to data privacy legislation at country level. If efforts are to be successful in globalising data, then more must be done to facilitate improved information sharing with greater consistency in terms of how related regulation is developed and applied.

“AI will undoubtedly play a more prominent role in supporting our industry. It will help deliver greater efficiency, reduce costs, and help remove many of the hurdles that exist in today’s customer journey. However, if we are to create a more productive role for AI then we must look to reassess our approach to data.”

FERMA Forum Today is in partnership with Captive Review, part of Newton Media.

Did you get value from this story?  Sign up to our free daily newsletters and get stories like this sent straight to your inbox.