Reuters event: Insurance execs see “huge opportunity” in AI, as well as liability challenges

AI was the dominant theme at Reuters’ The Future of Insurance USA conference in Chicago this week, with executives excited about the potentially transformational efficiencies and confident that it will help – rather than replace – their workforces, but concerns remain about potential liabilities that could arise.

On a panel discussion titled “Driving Forwards in an Ever-Changing World”, Louis Gritzo, chief science officer at FM Global, said that in the near term “there's massive efficiency through digitalisation and use of AI”.

“We put that in process a number of years ago, and we're starting to see the return on investment in that,” he said.

“We're not reducing headcount, what we're doing is we're optimising our headcount. We're making our people more efficient at the jobs they're doing, taking away the mind-numbing stuff, and allowing them to focus really on the needs of our clients.”

Gritzo said he considers traditional AI and Gen AI to be “very different”.

“We have used traditional AI in many different ways, shapes and forms over the past 10 years or so. A lot of the tools we give our clients are AI-based – a lot of our climate sciences have AI aspects to them – and then we have AI that we use in our own business processes to improve their efficiency. So those are fairly well established.”

Gen AI is a “big wildcard”

“Gen AI is the big wildcard right now,” he said.

Gritzo said FM Global is looking at using Gen AI to take the data that's available for natural hazards and fire, which are still the number one source of property loss for companies, and combine it with physics-based model calculations as an additional source of information. This will allow the insurer to develop innovative, cost-effective solutions faster, he said.

“So we believe that's a huge opportunity for us and our clients to do innovative things at scale and at speed that we've never been able to do,” he said.

On the same panel, Kristin Towse, executive underwriting officer at Markel, added: “From a generative AI standpoint we view that probably the best immediate use case for this from an efficiency standpoint is to help ease some of the administrative burdens and help knowledge workers in the digital age be able to focus on more strategic things.

“We're thinking about using it and implementing it more across the platform, to underwriting, to claims and certainly to the engineering side.”

However, Towse noted that the other side to Gen AI is that it “creates some interesting risks for insurance”. One example she gave is increased cyber exposure and the potential for bad actors to use technology malignantly.

Matt Kramer, US CEO at Ascot, highlighted that there are concerns among underwriters that AI may be a threat to their jobs. But he said that AI is actually a help to them.

“When we think about integrating AI into our everyday lives, of course the first response you get from an underwriter is, ‘Oh, boy, I'm gonna be replaced.’ But that's not it at all. It's there to aid the underwriter in making better decisions, intake of submission volume, scraping that data, collecting that data, using that data to our advantage and closing that information loop where that data can be used by everybody within the organisation.

“That's what AI can help us with. It's finding the efficiencies, it's collecting the data to make better decisions. It's not there to replace us.”

Kramer commented that insurance remains a people business, but AI can help bring in efficiencies “to help us get more quotes out of the door and find more quotes that are within our appetite”.

The executive also agreed with Markel’s Towse about the unknown liabilities that could emerge from AI. He gave a recent example of a lawyer using Gen AI to write a brief that cited cases that did not exist.

“This is a new challenge for us and I believe it's going to be regulated at some point, but it has to be something that integrates within our everyday use to help mitigate risk and to help us be more efficient and literally get it to the 21st century,” Kramer said.

Claire Burns, chief marketing and communications officer at The Hartford, also suggested AI would have a big impact on her company.

“We don't believe it's a question of if, but of when it disrupts everything that we do,” she said.

“We're very focused on making better decisions faster and on all aspects of the value chain, starting with underwriting, claims, software engineering, and then also investments, which is another big use case for how we can make better decisions faster.”

She also highlighted the potential adverse impacts from AI.

“We always look at both sides of the coin. So we're looking at: how can AI become an existential threat to our business? We're a big workers' comp carrier. If the predictions are true that AI is going to replace 20 percent to 40 percent of knowledge workers over the next five years, we're looking at that as a potential decline in revenue and we're thinking about ways we're going to offset that.”

AI to be “table stakes” within four years

AI was a dominant theme on other panels at the conference as well.

During an insurtech panel, Hippo president and CEO Rick McCathron said that AI is still in an early stage but moving fast.

“Whether it's from a customer perspective, from an internal expense control perspective, or from a pricing and then risk allocation perspective, I think we're three four years away from it being table stakes,” he said. “Then everybody is doing it, everybody has to do it, for you to be competitive at all.”

McCathron suggested that insurtechs may have an advantage in this.

“The big difference between insurtechs and legacy carriers is the rate and speed of change is night and day,” he said.

As an example, he said that Hippo in the past two years has been able to file and implement over 200 rate filings with an actuarial team of six.

“There’s a lot of technology that supports that in the back end,” he said. “The rate of speed is changing in the industry, and the companies that can keep up with it are the ones that are going to be successful long term.”

A panel on claims also discussed the impact of AI.

Charlie Wendland, head of claims at Branch Insurance, commented: “The biggest game changer we've had is just the integration of AI to free up our adjusters off administrative tasks so they can focus on more complex items or customer care. Large language models have been a true game changer.”

Jaime Palumbo, vice president of claims at Corvus Insurance, highlighted the speed at which the insurtech is able to use AI to connect policyholders to alert them about certain threats in the cyber landscape, which she said “has been really crucial”.

Eric Marler, head of claims at Counterpart, said his company is using AI and data to streamline processes, and “take away a lot of that administrative work and really make it so that when my team gets the claim assigned to them, they kind of have a running start with it”.

Dustin Angle, director, analytic solution advisor at Nationwide, said his company is “pretty far along in our AI journey at this point”.

Angle said a large piece of the work has been focused on building a larger ecosystem of models. As an example, he said AI can be used to allow the adjusters to focus on the more complex pieces.

“We have about 20 percent of our personal lines auto claims go directly to the repair process,” he said. “So all those traditional investigative pieces and looking at the point decisions that they're making, we have individual models for that and that's a huge piece for the customer. They're reducing that cycle time by two to five days on average for all those claims.

“People don't care necessarily about whether you are investigating the claims. They just care about their car getting fixed.”

AI accelerating the pace of decision making

Also at the conference, Zeev Avidan, chief product officer at OpenLegacy, a company that focuses on microservices-based API integration for back-end applications, commented: “You don’t want to be the only insurance company in the world that doesn’t use AI. Everyone will have to be in the race.”

Speaking alongside him, Aarti Gupta, head of insurance GTM – NAMER at Amazon Web Services, discussed how AI can be used to boost productivity.

“Let’s be real – the average age of an underwriter right now, it's about 55. Nobody is looking to get into an underwriter profession. So whether you're trying to maintain that tribal knowledge, onboard the new underwriters or really expedite that decision-making so nobody has to go through pages and pages of documents, it is really about synthesising that information but leveraging the cognitive skills of your underwriter for the decision-making process, because this is a regulatory environment.”

Gupta said it is not about giving the decision-making to AI.

“We are just really accelerating the pace,” she said. “The same thing goes with the claim adjusters and claims onboarding as well, and then more importantly, it's about elevating the customer experience.”

AI is also on insurance regulators’ minds.

At the conference, Adrienne Harris, superintendent in the New York Department of Financial Services, said: “We just put out proposed guidance for AI focused exclusively on pricing and underwriting. So we left for a later day claims administration and customer service and other aspects of the business.”

She said that a key takeaway is that no matter what the technology is, the principles of anti-discrimination continue.

“Just because we now have a fancy new technology, generative AI and other things do not mean that those age-old regulatory and public policy principles evaporate. In fact, it’s probably more important that we're mindful of those things,” she said.

The new guidance stated that insurers must do outcomes testing, in order to understand how the algorithm works.

“We really want our insurers across lines to be accountable,” she said.