5 key insights on analytics and AI in insurance
November 28, 2018 5:05 AM
Digital Insurance's Insurance Analytics and AI event brought together experts and practitioners in applying advanced data technologies to the insurance sector. Following are some insights from the keynote speakers.
Michael Costonis, EVP and COO for CNA, kicked off this year's event by talking about the changes coming to the industry as data proliferation leads to near-perfect information. "When we think of the data stream that exists in the world and our ability to harness that, we can really do some amazing things as an industry," he said. Rather than being an existential threat, analytics and data can help carriers "access customers we've never had before." At the same time, he noted, "we've got some systems that belong in the Smithsonian," creating room for AI and other automation technologies to help free data from legacy systems and put it in a more consumable form.
Colleen Saunders-Chukwu, AVP of data analytics and decision sciences for Nationwide, discussed the multiline carrier's shared-service approach to data strategy. IT has tended to be "capability-first," trying to find a solution for technologies, while the business has gone in the other direction, looking for a capability to solve a business problem. "That leads to solutions being brought up in the business on their own," she says, noting that Nationwide found that "we had multiple contracts for the same tool in some cases." Saunders-Chukwu outlined Nationwide's structure around enterprise data and analytics offices, which collaborate to get a view across the business of all the data assets.
Jerry Gupta, SVP of Swiss Re, talked about the ethical considerations coming from automating and turning decisions over to data-driven models. "Data scientists are creating algorithms that have a life-altering effect" making decisions about loans, insurance, banking and more, he notes. "If you have life altering system you need a code of conduct." Gupta suggested a "NEAT" structure: Models should be neutral, explainable, accountable with an audit trail, and transparent with their principles and values disclosed.
Pete Johnson, SVP for MetLife, and Jonathan Silverman, director for Microsoft, took the stage to talk about how the companies work together on a cloud infrastructure for MetLife's analytics initiatives. "Cloud is where AI happens -- you can't do things on-prem and maintain the infrastructure as more people start using it," Johnson said. He referred back to Gupta's talk and identified how the cloud can also help with some of the ethical considerations related to the tech. "Some of the solutions we see at places like MIT allow you to use advanced analytics without copying data at all," he notes. "You use blockchain as way to ensure that people are working with the same technology."
The morning concluded with Thomas Olds, VP of enterprise analytics and Nurur Rahman, data scientist (pictured) for Guardian Life. They talked about the need for a master technology solution to get full value from analytics. "The lack of true automation means scaling analytics becomes challenging," Rahman said. "Open source software, like Python and R, can help insurers automate and operationalize analytics and AI."