The latest technological advancement to represent a ‘game changer’ within many industries is the application of artificial intelligence (AI) – particularly natural language processing (NLP). While these capabilities are not new, they are becoming significantly more advanced and we are starting to see some interesting developments in the life insurance space.
At its core, NLP is the ability of a computer to understand human speech or text and respond in an appropriate manner. Chatbots are prime example of NLP in everyday use. Most people have experienced Siri on Apple’s iPhone or endured the voice enabled options when calling the contact centre of any of the big banks. While this technology has been used by the financial services sector for more than a decade, the degree of sophistication and adoption is increasing rapidly.
As an example, in January 2017, US-based general insurer GEICO released a virtual personal assistant called “Kate” that uses a natural language processor to respond to spoken or written questions from customers about policy coverage, billing information and insurance. Here at Bravura Solutions, our innovation lab is also researching this area and has been testing the use of Facebook chat to enable customers to query policy information held on our Sonata policy administration system, as well as submit claim notifications.
Chatbots present significant opportunities for forward-thinking life insurers and their customers. Applications along the lines of GEICO’s “Kate” are unsurpassed in their ability to instantaneously search vast document libraries and provide customers with immediate and highly accurate responses to their questions.
It’s not difficult to imagine how NLP could be employed by life insurers. Having extensive libraries of product disclosure statements (PDSs) and policy documents that cover products over many decades of business, makes it virtually impossible for a life insurer’s call centre staff to be familiar with the entirety of policy terms – particularly those who are less experienced. Today, complex customer queries are typically referred for investigation and response by specialist staff, which can involve considerable delays and sometimes human error. However, NLPs have the advantage of being able to instantaneously consider policy terms at the granular level and identify all relevant terms applicable to a customer’s individual circumstances, even those ‘hidden’ in the small print which may be easily missed by someone without significant familiarity with the material being reviewed.
Indian life companies are leading the way in the use of chatbots to enhance customer service and secure new business. In March 2017, India’s HDFC Life – in collaboration with chatbot platform Haptik – launched a chatbot that acts as a financial guide or virtual agent to help users choose the most suitable life insurance plans and solutions. Other life insurers in India – Birla Sun Life Insurance and PNB MetLife India Insurance – have followed suit and are using bots for customer support.
NLP is also being used by life insurers to drive efficiencies, particularly in claims administration and management. While many life companies have employed underwriting and claims rules engines for some time, these capabilities have been restricted to applying those rules to structured data. New NPL applications are going a step further by delivering life insurers the ability to analyse unstructured data such as complex financial documents and medical reports.
Early adopters are already reaping the efficiency dividends. In January 2017, Japan’s Fukoku Mutual Life spent $2.36 million to install the IBM Watson Explorer system to carry out tasks previously performed by claims adjusters. The system extracts data from scanned hospital records and medical certificates to determine claim payments, factoring in medical histories, injuries and procedures administered. And while these calculations are still reviewed by a person, the system has resulted in the redundancy of 34 staff, saving Fukoku Mutual Life $1.65 million per year in wages.
In March 2017, Europe’s fifth biggest insurer – Zurich Insurance – introduced AI to handle personal injury claims after a trial saved the company 40,000 work hours and dramatically reduced the time it took to review a medical report from an average of one hour, down to just five seconds. By eliminating the manual inspection of long and complex medical files, the AI system not only saved the company time, it also reduced claims leakage resulting from human error.
While the automation of manual tasks by AI and NLP capabilities is likely to result in staff reductions – as was the case with Fukoku Mutual Life – human interaction in the life insurance industry will remain, at least for the foreseeable future. Interestingly, when Japan’s Dai-Ichi Life Insurance recently introduced a Watson-based system to assess payments, unlike Fukoku Mutual Life, it did not cut staff numbers. The opportunity exists for life insurers to instead reassign those staff no longer required to areas better suited to their human capabilities.
Free from the repetitive administrative tasks now performed by AI systems, claims staff will be able to use their additional capacity to focus on the human dimensions of their role, such as providing personalised one-on-one support during the claims process and providing an empathetic voice to discuss their claimants’ health issues. Not only will this benefit claimants, it will also increase staff engagement by enabling them to focus on the more rewarding and fulfilling duties of their roles.
The application of AI and NLP in the global life insurance arena is unfolding at a rapid pace with the above covering just a few of the many examples now being seen around the world. From an Australian perspective, it’s now a matter of watch this space. It’s likely that domestic life companies are already experimenting and testing the waters with AI and NLP in an attempt to stay ahead of the game. Given that a number of the global life companies pioneering the use of AI in life insurance have Australian subsidiaries, the continued success of these AI strategies may well see this technology rolled out in Australia in the not too distant future.