As evidenced by the recent popularity of generative AI tools, the tech landscape in insurance seems to be moving at an increasingly fast pace.

“The AI space as a whole is moving faster than I think we’ve ever seen any technology move, ever,” said Jason Kolb, founder and CEO of InsurTech Dais. “In the 20 years that I’ve been in technology, I’ve never seen anything move this fast.”

But how can insurers weigh the challenges against the opportunities? Anand Rao, global AI lead at PwC US, said that this can be a tricky balancing act for insurers as, in some cases, technology seems to be evolving faster than the time it takes to even implement it into an organization.

“I think what’s interesting about the dynamic from the technology side is things are moving much faster than the way in which insurers are able to cope,” he said. “All these things are coming one after the other. It’s not that the insurers are not acting but that the pace at which they’re acting and the pace at which the technology is moving is sort of difficult to cope — not just for insurers but for anyone, for that matter.”

Kolb and Rao were speaking alongside Marie Carr, global growth strategy lead at PwC US, and Richard Clarke, chief insurance officer at Colonial Surety Company, on Carrier Management’s latest edition of Between the Lines, a video series that explores the topics in the latest magazine.

Rao said that with AI chatbots, in particular, the technology can teach itself to continuously evolve, which is adding to its fast development. While this can be beneficial to insurers using AI to assess things like loss reports and sort through data, one big challenge is ensuring the technology remains free from bias.

“Just the way in which these AI models are being built, you need to be very careful with how they are being built, because there could be bias that creeps into some of the models that you are developing,” he said.

He gave an example of using AI to assist in developing flood models.

“Let’s say you’re developing a model for flood in one particular area, so let’s say the Midwest,” he said. “There are specific parameters that go into it that may or may not really be applicable if you are looking at flooding in Florida or some other state. So, you really need to be careful on how you are using the data, where there could be a potential for bias and what kinds of decision you’re making.”

Another challenge with AI chatbots in insurance relates to information privacy, particularly when it comes to cyber policies, Clarke said.

“I think that issue of privacy is going to become a big issue,” he said. “The biggest source of insurance coverage for a commercial organization of any size, really — a large one, a small one — is going to be the cyber insurance policy because the core insuring agreement in most cyber insurance policies is breach of privacy or breach of network security, something along those lines.”

For cyber insurers that are already working through evolving risks related to ransomware, social engineering fraud or other exposures, AI chatbots are another one to add to the list, he said.

“[They] are now going to have to adjust and adapt the insurance coverage they provide for breach of privacy to make sure that they aren’t taking on risk that’s unnecessary,” he said.

As a result of this, Clarke believes insurance coverage will evolve along with AI technology. He said he sees coverage adapting with restrictions or exclusionary language related to AI chatbots in the future.

“I think absolutely there’s going to be an evolutionary aspect with respect to insurance coverage,” Clarke said. “Sometimes over the years, we’ve said that insurance carriers are slow to react, but I think with the rapid rise of this particular exposure, we’re going to see insurance carriers that will react rather quickly. I think we will see this restrictive language emerge if it’s not already in its infant stages.”

Moving Purposefully Toward Tech Enablement

Despite these challenges, Carr said that, overall, she believes carriers are doing better than in the past at adapting along with the technology and recognizing it as an essential part of their business.

“What I think is different now is that, if you look back over the last decade or even more, insurers have known that they’ve needed to become more digital. They’ve known they needed to become more tech-enabled, and they have started kind of making the investments,” she said. “What often happened is when they were laying out their strategies, some of the bigger things especially related to data analytics — the harder things — would kind of float to later on after they did the easier things.”

Carr said this has changed, however, particularly as the COVID-19 pandemic left many insurers scrambling to embrace more technology.

“I think what you see in the leadership of carriers is a realization that the tech enablement is necessary. Now, they’re moving much more purposefully toward it and not saying, ‘It’s too hard. It’s too risky,'” she said. “It’s like, look, this is where the world is going. We were already planning to go there, and guess what? There’s a lot of change that has come. We want to take advantage of that, and we want to kind of lean into that. So, I think that what’s different is the clarity of purpose of really being deliberate about how they’re investing and making it a strategic part of what they do.”

Clarke agreed, adding that insurers today are more equipped to keep up, and at a much faster pace than in the past.

“I think that it’s going to have a relatively rapid evolution,” he said, adding that he believes certain parameters will be set in place by regulations and litigation that set guidelines for intellectual property and information privacy. “I think once we get past those hurdles, [AI chatbots] will be used rather rapidly in the insurance industry.”

Use Cases for AI Chatbots in Insurance

Clarke said he sees many use cases in insurance for AI chatbots from an underwriting perspective, from correcting existing policy language to completely rewriting policies when necessary.

“I think there’s a lot of potential for artificial intelligence there,” he said. “Rewriting policies used to be, historically over the last 40 or 50 years, a massive project for an insurance carrier. It was kind of an excruciating process. It took a lot of time. You had to legally test the language and everything else, but I think now, you can do that much more easily, much more frequently with some realistic human oversight.”

Kolb sees opportunities for generative AI in underwriting, too. In fact, he had been following the evolution of this technology for years when he decided to develop his own tool with his team at Dais.

“I personally have been kind of fascinated with AI probably going on two years now since they first released some of the initial large language models,” he said. “It kind of became apparent to me the potential there. So, we’ve been kind of in exploratory mode with the technology for going on two years now, and then we started working on some ways to integrate this into the underwriting process.”

In February of this year, Dais, along with InsurTech digital brokerage The Paladin Group, unveiled UnderwriteGPT — a generative AI underwriting tool for the insurance industry. Based on large language models and generative AI, it’s designed to streamline the underwriting process and improve risk assessment.

“Around the time when OpenAI released ChatGPT, we were kind of on the tail end of the initial development. And so, the timing worked out well,” he said. “Especially with these large language models, you can do new things that simply were not possible before.”

With this in mind, he said UnderwriteGPT is already beginning to scale beyond underwriting to assist with other areas of insurance as well.

“There are so many really good applications within insurance, and we’re really just scratching the service with underwriting,” he said.

Dais is seeking to expand the tool into sales and customer service with coverage gap analysis, claims servicing, pre-renewal touch-ups and more. In fact, Clarke said he sees the use of generative AI overall creating a new cottage industry for third-party vendors that insurers or businesses, especially small businesses, can rely on to get their hands around using this technology.

“I think that small businesses are just going to have to ask for help. I don’t think they’re going to have the individuals, the means, the resources to do this on their own,” he said. “I think that gives rise to a new business model of outsourcing third-party contractors who make themselves available for hire for businesses to help them.”

Hands-On Experience, Adaptability

Kolb added that this type of hands-on experience will be vital for insurers to grasp the use of this technology.

“I think it’s very important to get hands-on experience with this technology because it’s very difficult to understand what it can do until you’ve actually had some experience with it,” he said.

He said he’s seen insurers already holding innovation events in which they’re able to scale their exploration of these new technologies as well as socialize them to the rest of the company. This is all part of adapting the technology to fit into their specific business models.

“What’s going to separate the winners from the losers down the road are the companies that start thinking about this earlier and start creating their own model based on their own culture, their own rules, their own processes,” he said. “They’re going to be positioned to scale that like crazy. Whereas companies that wait months, years, decades, however long it takes, it’s going to take a long time for them to catch up.”

Carr said insurers will need to realize that tech enablement is here to stay and the speed of it is going to continue to accelerate. She also said it’s important for insurers to be okay with making mistakes at first, as long as they don’t shy away from new tech ventures completely.

“Some of the best examples that I’ve seen are really leaders that have kind of made that shift where they’re like, ‘You know what? We’ve got to go for it. We’re going to make mistakes. We’re going to watch ourselves. We’re going to be wise. We’re going to adjust when we need to,'” she said. “We’re going to realize the technology and the rate of change is coming, and we’re going to make the best decision that we can make right now to move us forward.”