Technology is making a ‘big difference’ to underwriting – so will humans remain ‘essential’?

The human element of underwriting will remain “essential” despite the ability to automate more processes through the use of artificial intelligence (AI).

That was according to Vipr chief executive Paul Templar, who told Insurance Times that it would be “very hard to teach” AI to understand what a good or bad risk would be.

AI has become a trend that is here to stay in the insurance industry, with more firms starting to explore its uses.

For example, insurer Zurich said it was experimenting with ChatGPT earlier this year (27 March 2023) as it looked at how it could use AI technology for tasks such as extracting data for claims and modelling.

Later in the year (31 May 2023), insurtech Artificial Labs also announced that it was using the tool as part of a pilot to assist underwriters.

Templar felt AI was useful to assist underwriters, but did not feel it could replace them altogether.

“The human element is essential in the underwriting process,” he said.

“There is a lot of tasks and processes that can either be automated or have systems to assist with that, but one of the key skills that’s going to be very hard to teach AI and systems is how to replace that human instinct as to what’s a good risk and what’s a bad risk.

“It may well come in time, but today, we are well away from that.”

Data focus

This came after MGAA chairman Charles Manchester warned that underwriting will “change beyond belief” at the MGAA Conference earlier this month (6 July 2023).

Manchester said the shift in technology had got “everybody interested” and that there would come a point when “we’re going to see some real changes”.

Templar agreed with this, highlighting that with there being “more and more data” now available to underwriters, there was more focus being given to the analytics of that data and using it to improve the book.

“There’s also ways that you can bring machine learning AI into that whole analytical piece to improve the insights and make it much faster to get the insights,” he said.

“Then, if you couple that with the journey away from spreadsheets to more system-to-system integrations, which is what we’re pursuing at the moment, it means the speed that you get data is significantly improved.”

Bordereau

Vipr is a software solutions provider that also offers end-to-end management for the insurance market.

One of its offerings is a bordereau management system, which standardises large amounts of bordereau data, importing and converting multiple data formats to a single data standard.

A bordereau is a periodical report that is prepared by an insurance company for a reinsurance firm – it details assets covered and claims paid.

Templar said that it used to be a case that this would be sent to an underwriter in spreadsheet form 30 days after a month ends – and then they might not look at it for another 15-30 days.

“So when they looked at the data, it could have been 90 days old,” he said.

“Then they are making decisions on data that is really historic. But, then if you fast forward to a world where we are via the application programming interfaces (APIs) that we offer, we can connect to systems and they can send data to us in near real-time.

“All of a sudden, an underwriter has got the ability to make decisions on data that is less than 24 hours old, versus data that is 90 days old. So, it’s a big difference.”

Mistakes

Asked what mistakes firms can make with data, Templar said that the job to maintain the quality of data was “probably underestimated”.

He highlighted startups as an example, saying that they could face a challenge if they do not think about what systems they are using early on in their evolution.

“When businesses start and receive small volumes [of data] to a degree, you can do a little bit of that manually and they can maybe have a master spreadsheet that all this data can go into.

“But very quickly, it becomes unmanageable.”

He added: “Businesses that have thought about systems day one in their journey have been very successful, [but] businesses that are somewhere along their journey and then are trying to retrofit systems, it becomes a little bit harder.

“Certainly for a startup, it’s worth considering options quite early on in the journey.”