AI and oversight: Navigating risks in insurance tech

Complimentary Access Pill
Enjoy complimentary access to top ideas and insights — selected by our editors.

Artificial intelligence use has been on the rise across the insurance industry over the last few years, from smart home security to measuring the risks of severe weather events. For experts in the property and casualty insurance markets, AI promises to bring significant innovation — and great risks if not used carefully.

In a report published this year by Digital Insurance on advancements in insurance claims technology, 107 leaders across the industry were surveyed on what is needed to help improve claims technology. Roughly 70% of respondents work in the P&C insurance sector.

Not surprisingly, lack of process automation is the biggest hurdle facing insurance firms, with 58% of all respondents and 56% of P&C experts in agreement.

AI was the technology most in use or accelerating in investment to help manage claims with 33% of respondents. Cloud technology and mobile apps/digital platforms were close behind with 27% and 25% of experts respectively using or advancing funding for the products.

Ken Hugendubler, principal at Baker Tilly, said in a previous interview with DI that he predicts M&A activity will increase across the insurance landscape next year as firms seek capital to invest in technology, AI in particular.

"Insurance has seen a technology revolution for the last 15 years, largely accelerated by insurtech," Hugendubler said. "Almost every insurance organization has recognized the need to automate manual processes and find ways to solve challenges surrounding unstructured data, but many don't have the capital to address these issues."

Read more: Property & casualty insurance trends to watch in 2025

These resources address only one part of the tech puzzle however, as firms must be careful to have proper guidelines and oversight in place to minimize the risk of misuse.

Following the recent assassination of UnitedHealthcare CEO Brian Thompson in New York, previous insights into the company's use of AI algorithms when handling coverage approvals or denials were brought back into the public spotlight.

A report published in October by the U.S. Senate's Permanent Subcommittee on Investigations found that the organization's use of "at least one type of predictive technology" led to "increased adverse determination[s]." One of the models considered at UnitedHealthcare relied on determining factors like "[a] service's current denial rate, its potential for what the company deemed 'Fraud, Waste and Abuse' and the amount of money that could be saved by automating an approval rather than expending the additional labor of human reviewers."

"Prior authorization requests for post-acute care services for Medicare Advantage enrollees were denied at substantially higher rates than prior authorization requests for other types of care," the report said.

Read more: UnitedHealthcare's AI missteps a cautionary tale for banks

Below is a compilation of expert insight gathered by DI on the pros and cons of AI's growing presence in the insurance space and how firms can both benefit from the technology while working to minimize its risks.

A person using a laptop and phone with several graphics, including one that reads IoT.
Adobe Stock

What does Gen AI mean for raw data across the insurance landscape?

Data is a crucial lynchpin of the insurance industry, influencing everything from risk assessment to policy pricing and handling claims. But when it comes to structured versus unstructured data, seasoned professionals are enlisting the help of AI to create order from chaos.

In an opinion article for Digital Insurance, Chaz Perera, co-founder and CEO of New York-based technology firm Roots Automation, said that despite industry standards for formatting information, files are often multi-page documents that feature "email attachments, handwritten notes, faxes and others," he said. This has limited how effective underwriting teams can be.

"Up to this point, technology has been unable to keep up with the explosion of data around insurance, largely because the tools used by underwriting and claims teams rely heavily on structured data to operate," Perera said.

Read more: The impact of Gen AI on the insurance industry's unstructured data

Young man holding virtual screen displaying "AI" and related icons
Tierney - stock.adobe.com

Effective AI oversight is necessary for future growth of Gen AI in insurance

Analysts are confident that generative AI use will continue to grow in 2025, as tech budgets jumped this year across the insurance industry. But to ensure that the technology expands safely, firms need to first have proper governance protocols in place.

Sharing his thoughts on the topic with Digital Insurance, Jaspaul Saini, principal analyst at Celent, said data loss and hallucinations are two core concerns among a host of others that can be dangerous if left unchecked.

"Responsible AI governance should be the foundational construct of any carrier's AI strategy and needs to be considered at each and every stage of their gen AI journey," Saini said. "This includes the initial definition of the gen AI vision, selection of the required technology platform and stack and the establishment of an end-to-end process for gen AI development and rollout."

Read more: Responsible AI governance foundational for scaling Gen AI in insurance

Retirees are advised to tap into their Roth accounts last to minimize hefty tax bills associated with 401(k) distributions.
Bloomberg News

Can AI help stem the outflow of knowledge?

Like many areas of the financial services industry, insurance is suffering from a talent shortage. But experts say that modern AI tools can help capture the expertise of seasoned professionals before they leave the workforce for good.

To start, having those key leaders interact with the AI tools through conversation can help the models begin to understand the thought processes of the staff member and create a library of knowledge that can be used across an organization.

Experts from the global consulting firm Capco told Digital Insurance this month that AI can go beyond documentation to help organize mentorship sessions, guide new hires through the first few months of their job and more.

Read more: Using AI to reduce key person risk in insurance

Insurance claim form
Kaspars Grinvalds - stock.adobe.

The double-edged sword of AI in insurance claims

Whether a firm is just starting out on its AI journey, or is deep in the weeds, smart AI use begins with a focus on ROI and starting small, according to experts from the Chicago-based compliance solutions providers Origami Risk.

A report published this year by Digital Insurance on advancements in insurance claims technology polled 107 leaders across the insurance sector. Seventy percent of respondents worked in the P&C carrier profession.

When asked what their organization's top three priorities for digital transformation were over the next 12 to 18 months, claims appeared the most frequently as a top-three area — at 15% for top one, 27% for top two and 19% for top three. Customer service and product development also featured at 50% and 36%, respectively, for appearing somewhere in the three leading ranks.

Read more: AI's impact on fraudulent claims

di-using-smartphone-iphone-stock-062321
Brendon Thorne/Bloomberg

How consumers are turning to AI for insurance shopping

For professionals with the Palo Alto, California-based insurance broker Jerry, incorporating AI tools into the organization's tech stack helped create a 47% reduction in customers skipping or declining to provide documents during the application process.

"The AI validator tool cut median customer response time (the time between when a document is submitted and when it is approved) from around 47 minutes to around one minute, for photo and document submissions," said Josh Damico, vice president of insurance operations for Jerry.

Damico said that when supervised by a human agent in the event an application was more challenging than the tool was capable of handling, the AI validator was "as effective and accurate as human agents," he said.

Read more: Creative uses of AI to improve insurance shopping

For reprint and licensing requests for this article, click here.
Artificial intelligence Insurtech Property and casualty insurance
MORE FROM DIGITAL INSURANCE