‘Impossible for a human’: Zurich pilots AI in core underwriting process

Zurich’s Australia and New Zealand arm has concluded a six-month pilot with the University of Technology Sydney (UTS) to test the viability of an artificial intelligence (AI) model in overcoming a major hurdle within the life insurance underwriting process.

According to Zurich, the AI models, developed in partnership with UTS Rapido, the university’s dedicated research and development hub for industry/academia collaborations, have enabled immediate underwriting decisions on applications that include mental health disclosures – and without the need for a doctor’s report.

Previously, a decision on this part of the application process would typically take as long as 22 days, with access to doctors’ reports recognised by Zurich as a “significant friction” point within the application process.

Zurich notes that around one-quarter of the life insurance applications it receives include a disclosure of a mental health condition.

The AI models, which draw on anonymised data from seven years of past applications, identified factors correlated to a mental health exclusion being placed on an applicant.

Among these higher risk factors include:

• A lack of participation in sports, pastimes or recreational activities;
• An occurrence of any ongoing health issues beyond mental health, such as chronic pain, cancer or heart disease;
• Time spent in hospital or having medical treatment during the past five years; and
• Having biological parents or siblings with a health condition, such as diabetes, cancer or heart disease.

Zurich confirmed that the AI models will be used in conjunction with human decisioning, verifying decisions made by Zurich’s underwriting team and the existing rules engines they use. This, Zurich said, aligns with its AI principle to ‘always keep a human in the loop’.

Last year, mental health was the third most common reason for claims – accounting for 20 per cent of the total, with more than $255 million in claims payouts – processed across Zurich’s direct, retail and group insurance businesses.

For some of Zurich’s retail life insurance products, mental health accounts for up to 35% of claims in some cases, becoming the first or second most common reason for claim.

John Kim, chief data officer, Zurich Australia & New Zealand said the AI model analyses thousands of data points, something “not be possible for a human to do”.

“This project is a key step in Zurich’s journey to being Australia’s leading digital insurer, and importantly, it allows us to provide protection to more Australians at a time when they need it most.”

Zurich Australia & New Zealand’s head of retail, Jacqui Lennon, said the models could also have applications beyond underwriting, extending to the insurer’s preventative health and wellness offerings.

“In future, and data-prevailing [sic], models like these could potentially also be used to analyse and improve the underwriting of conditions beyond mental health, such as cancer, cardiovascular or musculoskeletal conditions,” Lennon said.