There is no doubt that the insurance claims process can be made more efficient – and artificial intelligence (AI) could be the best approach to streamlining a typically stressful procedure for both insurers and their customers. But experts caution that while AI does have smart applications in claims, it opens the door to a new set of issues that companies will have to tackle.
Lemonade, an insurance start-up based in New York, has been utilising AI to process claims. In one notable example, the company paid a claim for a stolen $979 Canada Goose jacket in just three seconds – something made possible thanks to AI.
CEO Daniel Schreiber explained that AI not only saves a lot of time, but can also help cut down the administrative costs associated with claims. The chief executive then mentioned that in the US alone, 11% to 13% of the premiums paid by customers goes to the bureaucracy of handling claims – something he calls “crazy.
“A bot handling claims in seconds delights customers and crushes costs,” he said, as quoted by The Financial Times.
AI can also be deployed to better detect fraudulent claims.
“There are a lot of opportunities in using AI to spot evolving fraud patterns,” Oxbow Partners consultant Chris Sandilands told The Financial Times, citing an example wherein AI can be used to detect personal links between people involved in the different aspects of a claim.
But there is a dark side to using AI, and some industry experts have expressed their concerns.
AI could make some types of claims fraud easier. For instance, people could use fake receipts to prove “ownership” of a stolen item, and it would be difficult for the AI of today to discern such details.
“Regulators are very wary of AI,” Cooley law firm partner Heidi Lawson commented to the publication. “They are worried about the financial stability of insurers because losses can be very high.”
But even without deliberate fraud, the AI of today still has trouble working out which claims should be paid in full.
During its first few years, Lemonade’s loss ratio was above 100%, which meant it was paying out more claims than it was receiving in premiums. Schreiber has attributed this to the company’s AI, and offered assurances that the systems have since been updated.
AI can also be used to skew claims in the insurer’s favour. The practice of claims optimisation, which involves paying out the minimum amount that a customer can accept without issue, can be made much easier thanks to automation.
Duncan Minty, an insurance ethics consultant, told The Financial Times that such a practice could enable a downward spiral of increasing fraud attempts. Customers that believe that their AI-powered insurers will not pay out the full cost of the claim could be driven to inflate the size of their future claims, Minty warned.