machine learning

German insurers follow global trend to AI

Robot hand holding lots of dollar notes
Robots pay quicker than bureaucrats. Source: Getty

The Bavarian dialect is difficult even for native German speakers in other parts of the country to understand, but IBM’s Watson artificial intelligence application easily deciphers calls from customers of Bavarian insurer VKB. (And emails too, obviously.) More responsive customer service is just one way AI is transforming staid German insurers as they undergo a digital revolution that is changing everything from how quickly claims can be paid to figuring out who is trying to defraud the company.

“Insurers, banks, financial services as we know them today won’t exist in 10 to 15 years,” Christian Rieck, professor at Frankfurt University of Applied Sciences, wrote in a recent book about robots in finance. “Whoever doesn’t wake up now will be forced out.”

US startup Lemonade uses algorithms to draw up new insurance contracts online in less than a minute, and has already attracted an investment from Allianz X, the digital wing of insurance behemoth Allianz. Chatbots with names like Maja analyze the needs of the customers. When it comes to claims, algorithms again decide on their merit, without any separate verification.

Alerting to unusual patterns

AI is also proving useful in detecting fraud. Ahead of the recent soccer World Cup, for instance, insurers registered an unusually high number of claims for television sets so that customers could buy a new set at the expense of the insurers. “In many cases, the description of damages does not match the usual pattern,” said Peter Holmstoel, head of the anti-fraud division at the German GDV insurance association. The industry group estimates the cost of insurance fraud at €5 billion ($5.7 billion) annually — a cost borne not by the companies but by other customers through higher premiums.

Now AI can sound the alarm when claims diverge sharply from these usual patterns. In one well-known instance, soccer manager Jürgen Klopp broke his glasses during a match in 2012 and numerous insurance customers used the pictures from television to claim their own broken glasses. AI today could immediately identify the fraud.

French startup Shift Technology has developed software that reviews claims for plausibility in what is called rules-based fraud detection. “It’s hard to fool our system,” claimed Bo Soevsoe Nielsen, head of central and northern Europe for Shift. “When new fraud patterns emerge, we can see that very quickly.”

No more poker face

Another AI application is upending health reviews for insurers. The Chronos product from US startup Lapetus Solutions has only to scan a selfie from an applicant, analyzing thousands of different points in the face, to determine age, weight, health status and even if the applicant smokes. This examination can tell the insurer more about the applicant’s life expectancy than any existing method, the company claims.

The insurance industry is investing more than other sectors into AI, according to Tata Consultancy Services, spending some $125 million annually against the average of $70 million.

AI may prove beneficial for customers but it will come at the cost of lost jobs in the industry as supercomputers take over numerous routine tasks. Robots can perform many of these simple operations more rapidly and cheaply, leading to a dramatic loss of jobs, experts predict.

But customers will benefit from much quicker service. At a recent event for a Frankfurt insurer, AI took less than five minutes to recognize a receipt from a cellphone photo and get it into the claim management system.

That Frankfurt insurer, Deutsche Familienversicherung, found only one service in more than a hundred it has on offer that wouldn’t be faster and cheaper with digital technology — a care package of bathrobe, flip-flops and reading material for customers going into the hospital will still be sent by snail mail.

Carsten Herz covers financial services for Handelsblatt. Darrell Delamaide adapted this article into English for Handelsblatt Global. To contact the author: herz@handelsblatt.com.

We hope you enjoyed this article

Make sure to sign up for our free newsletters too!