In a recent interview with Reinsurance News, Adam Denninger, Global Insurance Industry Leader at Capgemini, emphasised the critical role of AI integration in underwriting practices, while acknowledging insurers’ likely cautious approach due to various challenges and risks.
Denninger asserts that adopting AI technology is essential for property and casualty (P&C) insurers striving to become underwriting trailblazers.
He highlights AI’s capacity to uncover insights that humans may overlook, stating, “it can find things that humans might miss or might not have thought of.
“The ability to do that better than your peers is a distinct competitive advantage, so you’re going to see a huge push.”
He elaborates, “Another thing you can do is map a specific risk against other risks that are similar inside of your existing book on a level that you yourself might not have thought of.
“A third thing you can do with it is look at highly effective underwriters and actually assess across those books at a very, very deep level for those same types of patterns, and try to understand what makes one underwriter more effective than another.”
An example of insurers embracing AI can be seen in the recent collaboration between specialist global insurer Hiscox and Google Cloud. Together, they have developed the first AI-enhanced lead underwriting model in the London Market insurance industry. This innovative approach combines Hiscox London Market’s advanced technology platform, Hiscox AI Laboratories (Hailo), with Google Cloud’s generative AI technology to automate lead algorithmic underwriting from submission to quote.
Another notable example is Swiss Re’s upgraded version of Life Guide, now featuring Swiss Re Life Guide Scout, a Generative AI-powered underwriting assistant. This tool empowers underwriters to make professional queries and promptly receive Artificial Intelligence (AI)-generated responses along with the source of information, all within seconds.
However, Denninger also acknowledges the risks and challenges associated with increased AI usage.
He suggests, “You can use AI to supplement an underwriter and make an underwriter more effective, and what’s going to happen is you’ll have much more intense competition between different groups, so your underwriters are going to be expected to do more. The expectations from brokers, customers, and others, are going to go up.”
Moreover, Denninger points out AI’s limitations, explaining, “AI will automate the mundane work and help find patterns that underwriters are struggling with, but a lot of other pieces underwriters will still do, such as all the human engagements. At some point a human needs to make a decision. There’ll also be situations that the AI can’t process, which humans will still need to do.”
Denninger emphasises the necessity of striking a balance between leveraging AI’s advantages and avoiding over-reliance, stating, “How do you not go too far and put yourself into a box where you’re trapped by what only the AI’s limitations are? Or how do you avoid staying too far on the human side and miss out on all the advantages? I think finding the balance is incredibly hard.”
Furthermore, he stresses the need for active monitoring by underwriters to ensure factual accuracy, given AI’s propensity for inaccurate analyses.
He explains, “We all know that AI can hallucinate, which is a very human term that we place on it. It’s just looking at data patterns and coming up with the next best logical word. From our perspective, we say, that’s not true. You’re telling me lies. You’re hallucinating.”
He further elaborates, “It is incredibly important when pricing a risk, or understanding what a risk looks like and deciding whether or not it’s eligible for coverage, to understand the facts and truth about that risk, so accuracy is non-negotiable.”
Integrating AI into existing systems presents another challenge. Denninger highlights the difficulty of incorporating AI into legacy platforms, given the need to determine the optimal operational approach and making substantial IT investments upfront.
To mitigate operational and change risks, Denninger recommends a gradual integration approach for legacy carriers.
However, he notes that larger carriers with greenfield opportunities have an advantage: “they have the capital and the experience to actually make it happen, and all the connections with agents, and everything else that they need to have an effective insurance company. But they also aren’t handcuffed by their existing systems and processes.”
Similarly, pure startups have the potential to excel if they possess the capital and expertise in house to do it correctly.
To aid re/insurers in mitigating these risks, the Association of British Insurers (ABI) has introduced a new guide to assist insurance firms in the responsible use of artificial intelligence (AI). This initiative aims to minimise potential AI-related risks while maximising the significant opportunities and benefits it offers.
Overall, Denninger emphasises that AI adoption in underwriting will be incremental, serving as a valuable tool for outperforming mainstream carriers, albeit at a measured pace.
He concludes, “It doesn’t replace the fact that you still need to be trusted by your broker, and that you still need to have a good pricing model. It can do a lot of things, but it’s not going to completely alter everything that we do in insurance.”
