Why the Biggest Risk in AI for Underwriting Could Be NOT Using AI for Underwriting

Between the years 1951 and 1954 a total of 18 organizations in the U.S. installed the first general-purpose computer, UNIVAC I. Of these businesses, three were insurers. (A fourth insurance carrier cancelled its order due to cost overruns.)

Why did insurers decide to “risk” being early adopters? The reason is simple: our business, by its nature, consumes and produces vast amounts of data.

Even at huge cost and with laughable limitations – the machines weighed about eight tons and operated entirely using vacuum tubes (necessitating another innovation, cold rooms) – the investment proved well worth it. Overjoyed by their per-calculation savings one of these future-forward insurers went back to purchase two additional UNIVACs in 1956. (Before long, other insurers jumped on the digital computing bandwagon to help them contend with a problem all too familiar to us today: an insurance talent crisis.)

There was obviously huge risk in spending that money. But the risk of leaving on the table a technology that offered a giant leap in processing capacity was equally massive.

Insurance's Insatiable Appetite for Data

Seventy years later, the insurance industry is no less hungry for data management technologies. In fact, it’s exponentially hungrier.

Evaluating risk, often with relatively limited information or highly generalized aggregate data, to generate quotes was a very human affair. Underwriters in the 1950s conducted much of their business in face-to-face meetings, or over long lunches. Today, markets are global; the types of risks we’ve been able to enumerate have grown immensely; and the variety of risk factors determining the fair value of a commercial insurance contract has multiplied.

Just like 70 years ago, there are new technologies that bring fulfillment of the promise to radically transform insurance within reach. Today, the revolution in data processing is ‘artificial intelligence’ – or, more accurately, large language models and their cousins, capable of understanding data in context, not just as a mechanical sorting machine like the UNIVAC.

Today’s AI-powered solutions deliver better processing of more granular data, which in turn means more accurate underwriting and claims management.

To put a finer point on things: we’re far beyond categorizing investment in AI for underwriting as a “risk.” The real risk is falling behind the competition.

At this moment, insurers using AI are reaping the benefits, such as up to 90% increase in underwriting capacity. Just as important, AI is freeing underwriters from manually reviewing documents, reworking incorrect data and other non-core tasks that consume up to 40% of underwriters’ time, which could be spent assessing and pricing risks and working with commercial insurance brokers.

As effects from external forces (e.g., inflation, returns on investments, natural catastrophe cycles etc.) intensify, insurers need every advantage to keep policies profitable. Standing in underwriters’ way are intense competition among insurers (many of whom have adopted AI or plan to), inefficiencies and combined ratios, which at the end of 2023 were around 104%, sustaining a prolonged hard market.

That’s prompted industry watchers to predict “a step change in productivity” (per McKinsey) as underwriting leaders adapt to new realities by looking to match investment in new technologies and processes to address long-standing. It’s not just the challenge of more unpredictable events – from wars to droughts and floods – or new threats such as cyber-attacks. The competitive global environment, customer expectations, and challenging economic landscape make evaluating, pricing and administering risk more taxing than ever.

More precise evaluation of those externalities – and the ability to knowingly take on more risk – is one obvious route for underwriters looking for a technology dividend.

Another is customer experience.

Broker relationships in the 1950s rested on personal relationships (and a fair bit of corporate entertainment by underwriters). They are still critical in 2024. But brokers' focus today is speed, accuracy, and clarity. That’s how they win their business. And it’s how underwriters win, too.

"Better risk evaluation and more accurate premium pricing"

That’s why so many underwriters are looking at AI technologies to accelerate their data-driven activities. It means they can deliver better risk evaluation and more accurate premium pricing, quicker, and with greater contextual awareness. And that frees up time to add value in the uniquely human activities – understanding broker and policyholder needs, building relationships, and innovating to beat the competition.

One other risk to bear in mind. Those three UNIVAC insurers in the 1950s? MetLife, Franklin, and Pacific Life. Plenty of other insurers made it happily through the 1950s and beyond without taking the risk on crude, expensive early computers. (Or, put another way, took the risk of not investing in one.)

Let this be a reminder that it’s never too late to deploy a new technology that offers vastly improved accuracy, increased productivity / capacity and enhanced customer experience.

What we’re seeing now is the rapid maturing of AI-based tools to deliver results that are yielding huge benefits to underwriters and their customers. We’re well beyond the UNIVAC stage already. The risk of being the underwriter that doesn’t invest is already growing.

‍

‍

Feeling like your underwriting operations are stuck in the mid-1950s? You’re not alone: according to an Accenture study, insurance is listed first in Potential for AI Automation among major industries. We invite you to learn how Roots Automation is making it possible for insurers to harness this potential to create real and lasting value.

Download your copy of our latest, white paper, "Transforming Commercial Underwriting with Digital Coworkers" for insights into obtaining and effectively deploying the tools and technologies that are driving today’s insurance technology revolution.

‍