$50 billion to $70 billion.
That’s the estimated amount of industry revenue generative artificial intelligence (AI) could unlock for the insurance industry, according to a McKinsey report from earlier this year.
“Almost all insurance companies are currently using AI,” says Bryant University’s Assistant Professor of Mathematics Jessica Zhai, Ph.D., who teaches Bryant’s “AI Application in Insurance,” course. “It’s an important tool that could help with tedious work like cleaning data and modeling.”
While McKinsey’s evaluation suggested that marketing and sales, customer operations, and software engineering dimensions would see the highest areas of impact, Zhai — who previously worked as a consultant of actuarial science and risk management services for EY, specializing in solvency assessment, actuarial modeling, and product design — notes that AI is fueling dynamic changes throughout the industry.
From companies like Lemonade that use AI to craft customized policies and pay out claims to organizations such as Verisk that launched AI technology to detect photo fraud in claims, the possibilities of implementation seem endless.
But as AI continues to be integrated into the insurance industry, it’s vital that humans remain in charge of making principal decisions, says Zhai.
Where AI thrives: claims and underwriting
If you’ve ever needed to make an insurance claim for property damage, medical expenses, a car accident, or any other type of incident, you know that the road to reimbursement can be cumbersome. With AI helping to streamline operational workflows, policyholders are now receiving payments much quicker, says Zhai.
For instance, if a driver is in an accident, they can take a picture of the damage and submit it to their insurance company for a quick AI assessment (followed by any necessary human review). Previously, policyholders had to wait for an insurance adjustor to inspect the damage and take photos that would be sent to the company where someone else would determine appropriate repair costs.
This process of assessing risk, also known as underwriting, is another area where AI is making its presence known within the industry, Zhai notes.
To make informed decisions around pricing, actuaries rely on mathematical and statistical techniques — known as modeling — to analyze and predict risk.
“If you go to an insurance company and ask, ‘Can I buy this type of insurance from you?’, they need to run some tests and evaluate you,” Zhai explains.
This evaluation is a time-consuming process and has traditionally required manual data collection, risk evaluation, and policy customization. Through the assistance of AI and its interpretation of structured and unstructured data, however, processing time is reduced.
An AI eye
Zhai adds that auto insurance companies, which currently appear to be leveraging AI the most within the insurance industry, are also investing in technology to help determine an individual’s premium. For instance, the Progressive Corporation, one of the largest auto insurers in the country, uses Snapshot — a small device that plugs into a car’s diagnostics port and records information regarding the driver’s behavior. This data is then fed into a predictive analytics algorithm that the company will use to decide if they will lower, raise, or maintain the customer’s premium based on their determination if the driver is risky, neural, or good.
That innovation has potential application off the road as well.
Zhai explains that Chinese healthcare insurance companies are building nursing homes for policyholders and offering integrated care packages that combine insurance benefits with on-site medical and nursing services, which could potentially reduce costs by streamlining the delivery of care.
It’s an area of future growth that U.S.-based insurance companies could consider pursuing, notes Zhai, and a prime opportunity for AI application.
“With the agreement of the customer, the insurance company can install this type of equipment in the policyholder’s home, so they could monitor the individual’s health condition and provide help if they have any personal issues,” she says.
Industry potential, considerations
But with potential advances, comes important considerations for the insurance industry, including data privacy and security, says Zhai. For example, if AI is collecting information for predictive analytics, to adjust premiums, insurance companies need to ensure they have secure guardrails in place to protect policyholders’ data.
In some instances, insurance companies are using AI to read through contracts to spot errors — a practice that could cause its own issues if not properly supervised.
“It seems a little bit risky because contracts are very important for the policyholders and the insurance company,” Zhai says.
She shares that while AI can create great models, the tech must rely on humans (usually the company’s actuaries, CFO, or CEO) to determine their assumptions or “parameters,” conditions that are critical for defining the model’s scope, limitations, and applicability. Human insight is also necessary for interpreting the model’s results and reliability, and to determine what modifications are needed.
“AI can give you some choices but not the final decision,” emphasizes Zhai.