IMPLEMENTING NEXT GENERATION UNDERWRITING

Automation and AI are bringing in new levels of consistency, scalability and speed to the underwriting function. According to Grace Flowers, Chief Underwriting Officer at Zurich Insurance Spain, there are already material improvements in risk validation, pricing and underwriting fraud detection that are having a positive impact on loss ratios.

‘There are also massive benefits in the simplification and efficiency of processes that doesn’t impact loss ratios but still goes into that combined operations space,’ she added. ‘An underwriter doesn’t want to spend an hour putting 50 premises into a rating tool. This is where we see a golden opportunity.’

Generative AI is in its element amid the vast reams of unstructured data that daily cross an underwriter’s desk.

‘It helps us not only have more information but also to better understand the information we have,’ said Beat Stalder, Group Chief Underwriting Officer at Helvetia, adding that a problem buried on page 73 of a report can be quickly surfaced by generative AI.

It used to be if we needed more evidence to complete an application in the life space then that could translate into hundreds of pages of medical records. Now with Large Language Models, we can put the relevant information at the fingertips of the underwriter so that can focus their time on the decision-making and adding value.

DYLAN ROWE
HEAD OF L&H AUTOMATION UNDERWRITING SOLUTIONS, SWISS RE

Nick Kitchen, Strategic Portfolio Director at AXA Commercial, said this capability was already proving to be a ‘game-changer’. Twelve months ago, when the issues with reinforced autoclaved aerated concrete emerged, putting buildings at risk of collapse, AXA Commercial used generative AI to analyse 70,000 engineering reports in a week – something that would have taken over a year to do manually.

‘We could then reach out and speak to customers to see how we could help,’ said Kitchen.

He stressed, however, that AI is no silver bullet – ‘sometimes there are other solutions to a problem that are a lot cheaper’ – and that human underwriters will continue to have role.

‘Underwriting is a mix of an art and a science and we need to have conversations around complex risk and work out the best solutions and find compromises,’ he said.

A homogenised industry, where underwriting decisions are made by ‘computer says no’ models, leaving some sectors struggling to get coverage, is in no-one’s interest, he said. ‘We still need an underwriter to be accountable for the outcomes,’ said Kitchen.

Even so, the underwriter of the future will need to confident and comfortable using these technologies.

‘When it comes to pricing and portfolio management in the retail space, we should be transparent with underwriting population that there’s a new benchmark for future needs,’ said Zurich’s Grace Flowers, adding that new hires in this segment of the market need to be ‘competent, confident and proficient’ with technical tools and predictive models.

Here’s Grace Flowers, Chief Underwriting Officer at Zurich:

For more complex underwriting, then the skill set will be around decision-making that is both more thoughtful and holistic yet also quicker than ever. ‘This is where technical training needs to put more focus,’ she said.

By 2030, underwriting will be more efficient, more streamlined and capable of making better decisions at speed. ‘I see discrete areas where that pricing, analytical portfolio management capability will be far more sophisticated, data orientated and analytical,’ predicted Flowers. ‘It will be self-evolving, with that virtuous circle dynamic whether it’s machine or people learning.’

Insurers are already encouraging their underwriting teams to experiment with generative AI, using inhouse models to build confidence and competence. The efficiencies are evident, and the removal of more laborious aspects of the role very welcome.

And perhaps most importantly, it’s equipping the profession with skills and knowhow to stay relevant in a fast-changing world, where from cyber to climate change, underwriters need to understand, analyse and assess fast-emerging and complex risks.