Suncorp tests Watson AI for accident liability

Suncorp has revealed it is testing IBM’s Watson software in its insurance business to determine who is liable for an accident, particularly where multiple vehicles are involved.

Chief customer experience officer Mark Reinke told an investor day that artificial intelligence could usher self-service into a part of claims processing that has traditionally required humans to make decisions.

“One of the reasons that self-service in claims has traditionally been quite low, you would say, is that it gets complex once you’ve got multi vehicles involved, particularly in terms of determining liability and obviously that’s important to how the claim proceeds,” Reinke said.

“We’ve been able to test and prove the use of artificial intelligence in the determination of liability and that - in the testing that we’ve done - has made a significant difference.

“What that means is we can take a customer through that process without them diverting out of the process at some stage because we needed you to divert to a human effectively to make that decision.”

Customer platforms CEO Gary Dransfield revealed the technical underpinnings of the test platform.

“That case is using IBM Watson for that liability determination piece, and to make it happen 15,000 claim files had to be de-personalised and loaded back into the system for Watson to learn from,” he said.

Dransfield said Suncorp hoped to make the customer experience for claims “quicker and better … in terms of getting to choose and be allocated to a repairer quickly”.

For Suncorp, it will also “lower claims costs”, Dransfield said, “because we can help to facilitate a pathing outcome to our preferred repairers to get both the quality and the cost outcome that we’d like to see for the customer.”

Suncorp did not say whether the system is yet in production.

However, it is hoping that data science and artificial intelligence will have a broader impact than just its insurance business.

“We’re reviewing and streamlining our support frameworks such as our conversation models, advice framework and consent models to help deliver consistently and improve the customer experience right across the board,” Reinke said.

“We’re also continuously looking for opportunities for digitisation or to use artificial intelligence to enhance our channel network, streamline lower value interactions for customers, and tilt resource investment to assist our frontline to deliver significantly better customer experiences.”

Last month, Suncorp’s digital insurer, Bingle, launched what it calls Binglebot.

Dransfield called it an “artificial intelligence style support platform that provides contextual answers to our customer”.

“It gives them an immediate response but it also learns from how the customer responded to us, and the nature of the service we provided,” he said. “So it becomes intuitive and self-teaching.”

Suncorp also hopes to use AI to better understand where customers are at in their lives and therefore what types of products or services they might be interested in.