The age-old insurance underwriting model worked well for a traditional environment of slower and more predictable change, with similar processes and risk evaluation methods across insurers.
Today, the past is no longer a reliable indicator of the future. The fact that the global average temperature for each of the previous 11 months was the highest in recorded history confirms that we’re in uncharted territory when it comes to climate change. New technology is empowering cyber-attackers in ways that we can currently only speculate. Lifestyle patterns and medical costs are rapidly changing risk exposure and indemnity costs of disability, workers’ compensation and retirement. Risk is intensifying and diversifying at an accelerating pace.
The devil, as always, is in the detail, and there are a lot more of them than before for underwriters to analyze. Weather, health and other risk data has proliferated enormously, and forward-looking analytical models have become a critical tool in making sense of catastrophe probabilities. In addition, there are many new sources of data that shed light on risk. Geospatial data has the granularity to allow underwriters to differentiate between, for example, two properties on the same floodplain but with very different risk exposures.
This abundance of data is impossible to input, assimilate and analyze manually, even with basic digital tools like spreadsheets. Even traditional risk information can be overwhelming, with the result that underwriters are unable to give submissions they review the attention they deserve. As a result, risk assessment is increasingly patchy and imprecise, pricing is inexact, and the process can be daunting.
This presents insurers with two problems. First, it is significantly more difficult for them to accurately assess and price risk. And second, with customers and brokers demanding less onerous application processes and quicker approval, carriers that still use traditional underwriting methods and outdated tools will find it increasingly difficult to compete. They also may struggle to attract and retain the scarce underwriters and actuaries who are vital to their business.
In fact, Accenture research ‘Why AI in Insurance Claims and Underwriting’ found that up to 40% of underwriters’ time is spent on non-core and administrative activities — an annual efficiency loss of between $17 billion and $32 billion. More than half (60%) of the underwriters surveyed believe that improvements could be made to the quality of their organizations’ processes and tools.
Potential to reinvent underwriting
Fortunately, generative AI has the potential to reinvent underwriting in two critical ways: by giving underwriters better insight into the risks their customers face, and by streamlining the underwriting process. Together, these hold the promise of a big improvement in profitability and competitiveness by enabling more accurate pricing, more efficient processing and enhanced customer servicing.
New solutions driven by generative AI can identify data anomalies and gaps (e.g. mismatches in location listings across submission documents) and can automatically trigger and draft communications with brokers and customers to resolve these issues and save time. AI is also able to rapidly identify similar risks on the carrier’s books and compare them with the risk being assessed. This helps set the price for the new policy, ensures compliance with the insurer’s strategy and underwriting guidelines, and improves the consistency of its book of business.
We’ve seen impressive headway in the industry already, with global insurers such as QBE turning to Gen AI to transform its underwriting function for better risk selection and to support rapid growth. QBE is using a Gen AI powered tool within its U.S. cyber insurance division to streamline the quote generation process, reviewing submissions and summarising the key points, speeding up the submission to quote process by 65%.
Others may soon follow, as 85% of insurance executives globally say their organizations plan to increase spending on generative AI this year, according to Accenture’s most recent Pulse of Change survey. Almost three-in four (72%) are confident that they have the right data strategy and core digital capabilities – including the use of structured, unstructured and synthetic data – in place to effectively harness generative AI.
And many fintechs and insurtechs are at the forefront of applying generative AI to assess underwriting risks and transform insurance operations. One such insurtech is Cytora, which has developed a digital risk processing platform for insurers, while others, like hyperexponential, are hoping to use generative AI to automate the collation, inputting and analysis of data, saving a big portion of underwriters’ time and improving productivity.
I believe generative AI will transform every part of the insurance value chain, especially underwriting and claims. The ubiquitous spreadsheet will be around for a long time to come, but for underwriters at least its days are surely numbered.
