From wildfire risk to fraud, AI and analytics reshape reinsurance: Howden

AI tools should improve both speed and discipline at the same time, and no trade-off is required. So says Nathan Schwartz, global head of analytics at Howden Re, who spoke to Monte Carlo Today about separating hype from substance when it came to analytical tools.

Key points:
Misconceptions persist about AI’s true role
Fraud detection offers clear industry-wide gains
Speed and underwriting discipline improve together

“AI has really come to the forefront of everybody’s mind in the last couple of years, since ChatGPT burst on to the scene,” Schwartz noted, and indeed, at Monte Carlo this year, AI is dominating both formal sessions and informal exchanges.

“But AI has been around for a long time: machine learning from decades ago remains incredibly useful today. We've been using AI in much of what we're doing for clients.”

For Schwartz, one of the greatest misconceptions is that AI simply means generative models such as ChatGPT. “It is that, of course,” he acknowledged, “but it’s also many other things. It’s all these tools we’ve been using for years.”

Another misconception? That AI can already replace humans. “Someday, maybe we’ll be at that point, but it will be a while. There’s always some human element that needs to come in and guide the AI. It can make a human much more effective, but it’s not yet ready to take over our jobs.”

That human element, Schwartz added, is often being shaped by the firm’s newest talent. “You can see the future of AI in our newest employees who are just entering the industry. Many of our best applications of AI come from them. We’ve always encouraged everyone to innovate and drive our analytics forward, and that culture will ensure we make the most of these new tools.”

Far from being theoretical, Schwartz said AI was already embedded across Howden Re’s client work. “When we process data for our clients on casualty or cat submissions, we’re taking internal client data and turning it into something consistent with what the reinsurance market wants to see. Gen-AI helps us with that processing.”

He pointed to specific tools such as CQ (Cat Quotient), which gives clients an instant estimate of catastrophe risk at any location – crucial when underwriters need to decide on a quote in real time. “Machine learning and gen-AI techniques are making those tools more accurate and useful for our clients.”

Other immediate areas of impact include parametric products, where AI supports the data matching required for non-traditional structures, and fraud detection. “Fraud detection is a clear win for everyone,” he stressed. “Fraud is an enormous problem in insurance. To the extent companies can fight it better, it makes insurance more efficient.”

Speed and discipline

AI might often be seen as a shortcut but Schwartz emphasised that the best tools need to enhance both speed and underwriting discipline. “As AI tools make you more efficient at your job, you will be able to improve both dimensions at the same time. You don’t want to sacrifice discipline for speed, but the ability to do both is going to make insurance and reinsurance companies win in a competitive world.”

For the upcoming 1/1 renewals, Schwartz highlighted Howden’s Blaze wildfire score as a timely example. Built in 2024, it blends hazard propensity with exposure concentrations, particularly at the woodland-urban interface. “That tool has been incredibly valuable in California, especially after the horrible fires this year, and we’re now spreading it across the US and the world,” he said.

“These tools are going to become like our mobiles today.”

Beyond nat cat, Howden has also built new profiling tools in lines such as D&O. “These kinds of analytical tools can help our clients identify what parts of their portfolio are driving the reinsurance costs they’re facing; their tail risk, or even their day-to-day losses,” Schwartz said.

Looking ahead, Schwartz likens AI’s impact on insurance to the arrival of email and personal computers. “When PCs first appeared on everybody’s desk decades ago, nobody knew exactly how things would change, but we knew it would be dramatic. AI and machine learning are like that.”

For Schwartz, the most transformative shift might be in ubiquity. “These tools are going to become like our mobiles today, which seem like an extension of ourselves. That’s how AI tools will become for insurance companies; an extension of everything they do.”

Already, AI is proving “a shockingly good search tool”, particularly for contract analysis where wording varies. It is also enabling new approaches to synthetic data creation and profitability profiling.

The direction of travel is clear. As Schwartz concluded: “The storing and use of data is getting so much more robust, it’s going to continue to accelerate as the tools to use it get better.”

Nathan Schwartz is a global head of analytics at Howden Re. He can be reached at: nathan.schwartz@howdenre.com.

For more news from Monte Carlo Today, click here.

Did you get value from this story? Sign up to our free daily newsletters and get stories like this sent straight to your inbox.