Swiss Re exec talks three stages of insurance AI maturity

Register now

If artificial intelligence is about connectivity, then data is the parent that feeds the model, according to Jerry Gupta, digital analyst catalyst at Swiss Re, who delivered the opening keynote at Digital Insurance’s Insurance & Analytics conference in Boston.

The insurance industry is steadily witnessing the effects of AI, which is here to stay, Gupta says. As an example, he notes the emerging technology has increasingly been used in distribution and claims straight-through processing in recent years. Carriers are now utilizing AI in product pricing and underwriting.

Gupta believes insurance is in the second of three AI implementation phases—that is humans working on enterprise IT projects and letting machines take over at a certain stage. This is way past the days where humans were in charge of both coding and execution, he says.

As carriers implement more AI into their systems to improve self-service claims, underwriting or market segmentation, it’s important to remember that IQ smarts aren’t what separates companies like Google or Amazon from the rest of the pack, he says: It’s their innovative culture.

“People need to experiment and fail and have innovation metrics as part of their core business,” said Gupta. “Your day job is not different from your innovation job. It’s the same.”

According to Gupta, the third level of AI implementation implies machines will do the heavy lifting. Machines are already teaching and communicating with each other, allowing insurers to receive unlimited amounts of data from Wi-Fi enabled devices, car sensors and biometrics. This trend will lead to the “Uberfication” of insurance where “no humans are involved apart from the developers building,” Gupta says.

Additional proof of this impending change lies in the amount of venture capital investment in AI startups. One in five startups in insurance today predominantly use AI-based services, according to Swiss Re research. However, 100% of insurtechs the reinsurer has encountered have some AI component in their product.

Full dependency on machines will happen in the next 25 years, Gupta says. But there are challenges. One roadblock is data privacy. There is little clarity as to what customer data insurers can use. Other challenges include system integration and brand impact—the fear an uneven customer experience thanks to greater dependency on vendors.

“It’s difficult to trust a third party to handle customer experience when it’s a core proposition of your business,” he concluded.

For reprint and licensing requests for this article, click here.
Artificial intelligence Machine learning Start-up funding Automation Data quality Unstructured data Customer data Data modeling Swiss Re