Seeking human empathy, health insurers turn to AI

When Florida Blue wanted its call center employees to demonstrate greater emotional intelligence when dealing with customers, the nonprofit health insurance company enlisted a tutor incapable of emotion.

A generative artificial intelligence, or genAI, chatbot instructs 30 Florida Blue customer service representatives on how to behave like human beings when interacting with other human beings. The chatbot guides workers on human behaviors, such as when to slow their speech, when to hasten a call to its conclusion and what to recommend to policyholders. The company plans to expand this pilot program to its entire 1,600-person call center team this year.

Related: Cigna's customer-centric strategy could ripple across the sector

“A lot of the time, people carry emotion into calls with health insurers. When you're upset, it just comes out. One of the prompts is to remind the advocate, ‘Hey, this member appears to be stressed. Make sure you're pausing and listening to them,’” said Anne Hoverson, vice president of digital transformation at Florida Blue, a subsidiary of Guidewell.

Insurance companies already used genAI for processing claims, predicting clinical needs and performing administrative functions, but this latest trend is different, said Josh Streets, a senior consultant at the International Customer Management Institute, which advises business on call centers.

Emerging technology such as genAI is now part of the customer service toolkit for insurers such as Florida Blue, UnitedHealthcare, Independence Blue Cross and EmblemHealth, which have embedded the software into mobile apps, virtual assistants and call centers.

The health insurance sector has always had a customer service problem: Consumers don't like being told no, and a core function of health insurance is directing — and limiting — access to care to constrain escalating healthcare costs for employers, government programs and individuals. The assassination of UnitedHealthcare CEO Brian Thompson in December and the lionization of his alleged murderer in some quarters provided a shocking reminder that Americans are deeply unhappy with their health insurers.

Yet the health insurance business model is incompatible with the underlying demand that health plans cover more things and refuse to cover fewer of them, which may limit the sector's ability to solve poor relations with members. Health insurers have made concessions on coverage in the recent past, though. UnitedHealth Group, Cigna and several Blue Cross and Blue Shield carriers have eased some prior authorization requirements, for example.

But insurers are trying other means to, at a minimum, make interactions between customers and company representatives less unpleasant. Cigna now links employee compensation to customer satisfaction as part of a broader campaign to improve service, for instance.

'More important now than ever'

In the aftermath of Thompson's murder, UnitedHealth Group CEO Andrew Witty pledged the company would strive to win public trust. "We understand and share the desire to build a healthcare system that works better for everyone. That is the purpose of our organization," he wrote in a New York Times op-ed in December.

For-profit UnitedHealthcare seeks to reduce operating expenses, save time and provide members with flexible means of communication with the company, said Hans Hage, chief product officer at the UnitedHealth Group subsidiary's consumer operations division.

The UnitedHealthcare mobile app, for instance, has grown increasingly popular, Hage said. Last year, 60% of members used the app to contact the insurer, he said. The app handled 107 million interactions, more than double the number in 2023, and nearly 100 million users downloaded it, he said.

Last year, UnitedHealthcare fine-tuned the app’s search capabilities. Users can type in a symptom, such as knee pain. The app then responds with potential conditions, treatment recommendations, available providers and estimated costs. It also reflects customer preferences, such as favoring a hospital outpatient department over an outpatient clinic, in the results.

That rising demand led UnitedHealthcare to incorporate genAI into its customer service platforms, Hage said. “It's more important now than ever that we continue to advance the transparency that our app allows consumers to embrace and really lean into member preferences to get the information in their hands,” he said.

Human UnitedHealthcare employees verify that the genAI is accurate and connects procedures to the appropriate billing codes, Hage said, which is standard in the industry. This may head off some criticism and skepticism after UnitedHealthcare and other insurance companies attracted negative attention for allegedly using AI to replace clinical reviews and to reject claims.

'Robots don’t have souls'

Therein lies the challenge with the technology, said Mike Vennera, executive vice president and chief strategy, technology and operations officer at Independence Blue Cross. GenAI is a promising but sometimes unreliable innovation, he said. Customers could feel frustrated if it produces inaccurate results, ironically leading to a worse consumer experience, he said.

“We didn't want to expose the generative AI model directly to our members," Vennera said. "We wanted to make sure we had an agent who would be able to see the result and, if it didn't look right, be in the loop.”

At the start of the year, Independence Blue Cross introduced a chatbot for its 700 call center representatives to use to look up benefit information for employer customers, he said. Those employees represent about 15% of the insurer's workforce. The Independence Health Group subsidiary hopes genAI will make the customer service system more efficient, Vennera said. But it does not anticipate the technology will reduce staffing needs, he said.

The industry appears split on whether genAI can replace human customer service agents. According to a survey results the International Customer Management Institute published in December, 35% of companies believe genAI will enable them to employ fewer call center workers compared with 38% who believe it will not influence staffing.

EmblemHealth views the relationship between genAI and humans as complementary, said Karen Ignagni, executive board chair. The technology cannot replace human call center workers, but it can help them improve their people skills, she said.

The nonprofit insurer employs natural language processing to advise its 400 contact center representatives on how to regulate their speech. This year, EmblemHealth aims to use genAI to comb through service call transcripts and identify trends in policyholder queries and responses, Ignagni said.

“Robots don’t have souls, but I do think there’s a lot to be said for humans sometimes needing to be reminded of how to communicate better,” Ignagni said.