One of the most frequently-used phrases at business events these days is “the future of work.” It’s increasingly clear that artificial intelligence and other new technologies will bring substantial changes in work tasks and business processes. But while these changes are predicted for the future, they’re already present in many organizations for many different jobs. The situation brings to mind the William Gibson comment, “The future is already here-it’s just not evenly distributed.” The job and incumbent described below is an example of this phenomenon. It’s a clear example of an existing job that’s been transformed by AI and related tools.
Kristen Buonopane and Haven Life
Kristen Buonopane loves her job as a “digital life underwriter,” though she says that neither she nor any other underwriter she knows prepared for the job as a child. She’s been an underwriter since she graduated from college fourteen years ago-for more than a decade as a traditional underwriter doing manual reviews of paper-based applications for life insurance, and then for the last two-and-a-half years as a “digital life underwriter.” That job came about as a result of MassMutual, her employer, acquiring a new business unit called Haven Life.
Haven Life is an “in house startup” offering term life insurance policies issued by MassMutual. The startup is only 4 years old, whereas its parent company, MassMutual, is almost 170 years old. The purpose of the digital agency was to create an end-to-end solution that makes it much easier to buy term life insurance. Here’s a short video about the application process:
The startup is autonomous for the most part, operating out of New York City. It has nearly 250 team members dedicated to creating new technologies, products and distribution channels, but draws on MassMutual for help with product development, data science and underwriting standards.
The role of underwriters is to assess applications for insurance-in Buonopane’s case, term life insurance. They decide whether to insure a prospective client and, if so, what premium should be charged for that customer’s level of risk. Broadly speaking, if applicants are expected to be healthy and live a long time, they are both more likely to be accepted as a policyholder and to be charged a lower premium. Typically-and in the case of Buonopane’s job working alongside the Haven Life team-actuaries develop decision rules and premium levels for different levels of risk, and underwriters apply them to particular client cases.
However, in the recent past the process of reviewing a prospective customer’s application and making an underwriting decision was not a smooth one. Buonopane and her colleagues had to read through a client’s entire paper application, try to identify the salient facts and risks in it, and remember or look up the relevant underwriting rules. In the case of medically underwritten coverage, all prospective clients were required to have a home visit from a paramedical examiner for a blood draw and physical examination. The process was both time-consuming and inconvenient for the client, and expensive and error-prone for the insurer.
The New Digital, AI-Based Process
Now, however, Buonopane and her colleagues do their work with a digital and AI-based system that only requires their attention where it is particularly necessary. The Haven Life Risk Solutions team, in partnership with MassMutual, has developed a platform that uses both a rule engine and machine learning models to analyze application and third party data in real time. It can now help MassMutual make many underwriting decisions without human underwriter intervention, and in some cases also without a medical exam. A recent enhancement to the no medical exam experience, internally known as LiteTouch, integrates a human underwriter review of the application in situations where minor data points might have caused an applicant to need a medical exam. With this enhancement, 21% fewer clients require a physical exam to get their policy decision. For those clients that still require an exam, a second machine learning platform is able to review medical exam results paired with applicant data to determine a final decision and rate. This model, created by MassMutual and utilized by Haven Life, is based on more than 20 years of data and 1 million life insurance applications. Ram Ballesteros, product lead for the LiteTouch initiative, said that half of the policies sold by Haven Life require no review by an underwriter at all.
When Buonopane does have to review an application, the process is generally much easier, although in some ways the actual decision is harder. She says Haven Life’s “platform” is very sophisticated and intuitive. She no longer has to comb through the entire application, which includes more than 100 individual data points; the rules and algorithms handle that. Instead, the system flags risk issues for underwriter review, and that’s all she needs to address in her workflow. The flagged data points commonly involve issues like a particular prescription drug that the applicant has been given, or something in the medical history.
In some cases, however, the decision is harder because she only sees the more complex risk issues now-the easy ones are handled automatically. Buonopane says that even though the decisions can be harder, it’s more fulfilling to focus on complex cases because it leverages more of her knowledge and experience. She also noted that the system is designed to dynamically filter all the applicant data and present only the risks that need evaluation. As she toggles between each risk, she is presented solely with the elements that are most relevant to make a decision. Ultimately, the system enables her to evaluate more complex cases in a highly efficient manner.
A typical issue she might have to address involves the reason for a particular prescription the client has taken. The drug Zofran (Ondansetron in generic form), for example, helps to prevent nausea and vomiting after chemotherapy or surgery, and is also often prescribed to women during pregnancy. If the client’s prescription record includes this drug, Buonopane has to determine the reason why it was prescribed. If the applicant is female with children born shortly after the drug was prescribed, she presumes pregnancy is the reason; if an oncologist prescribed the drug, the applicant is presumed to have had chemotherapy for cancer. Specific and common instances like this one with prescription history are why the LiteTouch initiative has been so effective in removing the need for a medical exam with some clients. Buonopane is able to clear up this data point before putting a client through a potentially unnecessary medical exam.
The Future of Human Life Underwriters
Both Buonopane and Ballesteros said that the AI-based system for making underwriting decisions is constantly evolving and able to address increasing percentages of applications without human review. I asked Buonopane, for example, whether she thought the system would eventually be able to determine the reason why an applicant was prescribed Zofran. “Probably,” she said. “But I think complex cases will always require a human review. Human underwriters can take a holistic perspective that a machine can’t.”
Buonopane didn’t have any formal training in computer information systems in college-she majored in business-and she says that tools like AI “weren’t even on my radar” when she started underwriting. But she says it’s now clear that AI and digital processes are where underwriting is going. Some underwriters don’t welcome this change, but she attributes her success as a digital life underwriter to always being willing to try new things and learn new skills. “Can we make it better and faster?” is a question that both the underwriters and the Haven Life Risk Solutions team are always asking, and they have a virtual meeting every week to discuss improvements in the system and process. “If you have a digital mindset you will be fine,” Buonopane said.
She also said that underwriting has traditionally been a somewhat isolated role, but is less so now. “I have always worked at home,” she said, “and before it was just me, sorting through several fragmented databases and relaying my decisions or questions to a broker or agent.” Now, however, she has direct contact with prospective clients via email if she has a question or needs clarification about their application. There are also the weekly, collaborative discussions with Ballesteros, developers, engineers and the other underwriters.
“Being able to help create the platform has been so rewarding professionally and personally for me,” she commented, “and I can incorporate personal service to the policyholder into the job.” She tries to customize the communication with the client, using empathy to make them feel that they are not just a number. “That would be hard for an AI system to do,” she noted.
Buonopane did have one concern about the future of life underwriting that the industry has yet to address. Digitization and AI are creating a greater need for experienced underwriters than for entry-level ones as the less complex aspects of the job are automated. She commented, “We’re not sure where the experts of the future will come from if we don’t need as many underwriters at the entry level. For our industry that’s a long-term challenge.”
Kristen Buonopane concluded our discussion by saying, “Nobody says, ‘Hey, I want to be an underwriter.’ It’s not for everybody-you either love it or you don’t. But I am passionate about analyzing risk, and love using my critical thinking skills. I am fortunate to be in a role that has made underwriting even more rewarding for me.”
Article by channel:
Everything you need to know about Digital Transformation
The best articles, news and events direct to your inbox
Read more articles tagged: Future of Work