Business News Daily: Artificial Insurance? How Machine Learning Is Transforming Underwriting

Adam Uzialko

Artificial insurance improves several insurer pain points while simultaneously benefiting the customer. Here's how.

  • Artificial intelligence (AI) can help insurers assess risk, detect fraud and reduce human error in the application process. The result is insurers who are better equipped to sell customers the plans most suited for them.
  • Customers benefit from the streamlined service and claims processing that AI affords.
  • Some insurers think that, as machine learning progresses, the need for human underwriters could become a thing of the past – but that day might be years away.
  • This article is for insurers, business owners and insurance company customers interested in seeing how AI benefits them.

Although it’s an industry that has proven resistant to change for centuries, insurance is undergoing a digital revolution. With the advent of advanced machine learning algorithms, underwriters are bringing in more information to better gauge risk and offer tailor-made premium pricing. On the back end, AI is streamlining the insurance process to connect applicants with carriers more efficiently and with fewer errors.

This rapid change means big things for insurers and applicants alike. Here’s how AI is on the frontier of the insurance industry and where it might be heading in years to come.

Assessing risk

Historically, insurance underwriters have relied on applicant-provided information to assess clients’ insurance risks. The trouble, of course, is that applicants could be dishonest or make mistakes, rendering these risk assessments inaccurate.

Machine learning, specifically natural language understanding (NLU), enables insurers to pore through more abstract sources of information, such as Yelp reviews, social media postings, and SEC filings, pulling pertinent information together to better assess the insurance carrier’s potential risk.

“Our ability to actually look at these textual data sources and pull out highly relevant information is greatly increased [with NLU],” said Andy Breen, senior vice president of digital at Argo Group. “We’re making use of these information sources that weren’t available or easily disseminated before.”

More accurate risk assessments mean more appropriate premiums. In an industry where the largest difference between insurance companies is not their products but their prices, a more individualized exposure model could make a big difference, said Sofya Pogreb, COO at Next Insurance.

“Traditionally, [the industry has offered] ‘lowest common denominator’ products: a standard liability policy,” Pogreb said. “What you end up with is a very undifferentiated product, where a bakery and a laundromat have the same policy. That’s not the right way to go for the customer. Being able to consume more data automatically, we will see more customization, and customers will benefit by paying for coverage they truly need.”

Detecting fraud

Fraud is a major concern for insurance companies, and AI is a key watchdog in the fight against fraudulent claims. As Samsung notes in a blog post about insurance fraud prevention, it’s all about detecting patterns that might escape human cognition:

“French AI startup firm Shift Technology incorporates this technology in their fraud prevention services, which have already processed over 77 million claims. The cognitive machine learning algorithms have reached a 75% accuracy rate for detecting fraudulent insurance claims. The ML algorithms provide details on suspicious claims with potential liability and repair cost assessments and suggest procedures that can resolve and enhance fraud protection.”

“The ability of machine learning to assist in spotting suspected fraud is well established, but human-led data science is just as capable so far,” said Areiel Wolanow, managing director at Finserv Experts. “The key difference over time will be one of cost. Professional criminals will keep abreast of industry-leading fraud indicators and adapt their behavior to suit. Human data scientists will need to iterate their analysis over time to keep pace, while machine learning algorithms train themselves over time based on observable changes in the underlying data.”

Reducing human error

The distribution chain in the insurance industry is winding and complex. A series of middlemen examine information between the insured and the carrier, leading to a lot of human error and manual work that slows the process, said Breen. However, AI is starting to fix that problem.

Algorithms can reduce the time and number of errors as information is passed from one source to the next. By logging in to a portal and uploading a PDF, the insurer reduces the amount of data entry and reentry and increases the accuracy, Breen said.

“People get tired and bored and make mistakes, but algorithms don’t,” he added.

For Pogreb, bridging the gap between the insured and the insurer is as important as reducing error. With better data, both customers and insurers benefit, she said, because insurers can develop better products based on more accurate assessments, and customers will pay for exactly what they need. [Read related article: Websites for Comparing Small Business Insurance Quotes]

“With machine learning, I think we’ll be able to do a much better job giving the consumer that advice automatically,” Pogreb said. “Based on what you tell me about your business and what I know about similar ones, [I can say] I believe this is the right combination of coverage for you. So it’s putting the onus neither on the agent nor on the customer – who frankly doesn’t have the experience or knowledge – but letting the data provide the advice.”

Customer service

Even in a sector as change-resistant as insurance, good customer service is paramount. After all, people often stop using companies with bad customer service. That’s why so many insurance company websites now include chatbots. These AI tools can guide customers through numerous queries without human intervention. They’re also available 24/7, unlike many teams of actual people.

For example, a customer who needs help accessing their account could ping the chatbot for assistance right from the insurer’s website. This function could potentially resolve customer crises in a jiffy. Real, human customer service agents may still be necessary for more complex concerns, but AI chatbots can handle most of the remainder.

Claims processing

Insurers exist to process claims and help customers cover them, but claims assessment isn’t easy. Agents must review several policies and comb through every detail to determine how much the customer will receive for their claim. That can be a painstaking process – and AI can help.

Machine learning tools can rapidly determine what’s involved in a claim and forecast the potential costs involved. They may analyze images, sensors and the insurer’s historical data. An insurer can then look over the AI’s results to verify them and settle the claim. The result benefits both the insurer and the customer.

Does AI in the insurance industry benefit the consumer?

Widespread industry adoption of a certain technology often reflects the benefits it offers to companies in the sector, sometimes with no obvious effects on the customer. That isn’t the case with insurance industry AI, which does have clear advantages for the customer.

AI-assisted risk assessment can help insurers better customize plans so that customers pay only for what they actually need. It can also minimize human error in the application process, so customers are more likely to receive plans that properly fit their needs. Of course, it can also expand an insurer’s customer service options and streamline the claims approval process. The end result is customers getting what they need.

The future of insurance AI

The insurance industry has only begun its foray into AI, and companies are already experimenting with new ways to incorporate it into their day-to-day operations in anticipation of further technological development.

“It’s the very early days of AI,” Breen said. “For menial, repetitive tasks, we put the computer on it … but we’re a ways away from a computer underwriter. We’re really just augmenting humans at this point.”

That’s still a significant change in the industry, he said. Underwriters at Argo Group are now beginning to manage portfolios, rather than review every single submission. The more standard, predictable claims are handled by machine learning algorithms, Breen said, and the human underwriter is essentially fine-tuning the entire process and intervening in cases that need higher-order decision-making.

Pogreb sees even more potential for streamlining the underwriting process. She expects that the number of applications a human underwriter is required to handle will significantly drop as machine learning finds its place in the insurance industry.

“We believe with technology and machine learning, a lot of [human underwriting] can be done away with,” Pogreb said. “The percentage of insurance applications that require human touch will go down dramatically, maybe 80% to 90%, and even to low single digits.”

While AI adoption has come in rudimentary ways, it’s already drastically changing the lay of the land. Insurance companies that want to stay competitive should test the waters of AI, Wolanow said.

“Companies can prepare and stay competitive by starting to assess the impact of machine learning on their business by prototyping their own algorithms,” he said. “An individual machine learning algorithm that performs its analysis on a stand-alone basis is actually quite inexpensive, [and] in many cases, a stand-alone analysis tool is more than fit for purpose.”

Max Freedman contributed to the writing and reporting in this article. Source interviews were conducted for a previous version of this article.

Adam Uzialko