ARTIFICIAL INTELLIGENCE IN LIFE INSURANCE INDUSTRY

Artificial Intelligence, popularly known as AI, is a very commonly used word for the last few years in the insurance industry.  As the name suggests, “Artificial” is not “Natural / Human”.  But it is important to understand that Artificial Intelligence is a human-designed modelling, it is not a different system in itself. It’s a logic used to learn from the past experiences of the available data and create intelligence therefrom because, by definition, Intelligence is the ability to acquire and apply knowledge and skills. Knowledge means facts, information, and skills acquired through experience or education; the theoretical or practical understanding of a subject. So Artificial Intelligence is about repeatedly and seamlessly aping human intelligence and knowledge based on experience and applying logic as it would have been applied by a human being.  There are various tools and applications of AI. Machine Learning (ML), Natural Language Processing (NLP), Robotics Process Automation (RPA), Speech Recognition (SR), Vision Systems (including Face Recognition) are all different types of Applications / Tools of AI.  Machine Learning (ML) is about an application/logic that can automatically learn on its own without being specifically programmed to do like that.  It learns from the past experiences and trends and actually improves with experience. AI and ML are the buzz words in the last couple of years and there are many myths and notions about AI, ML etc.  Though ML is an application of AI, which is technically nothing but acquisition of knowledge or skill, it is very common and often used as a conjunction or conjoint of AI in normal communications. While the words analysis and analytics are also interchangeably used, there is a wide difference between the same.  Analytics is not same in any way as a trend analysis or analytical reports generated through spreadsheets.  Analytics has the ability to use all available data including unstructured information and putting it in the right context of prediction.  Also, unlike spreadsheets and normal SQL queries, it has an ability to analyze a large number of variables. The biggest advantage of AI is technically customer one view and avoidance of multiple data versions. AI is going to be future in many industries.  The future of the Insurance industry will be heavily dependent on Artificial Intelligence in the years to come.  AI will play an important role right from business planning to product designing to customer acquisition to customer onboarding to servicing to claims & risk management.  In fact, the application of AI is already begun both in the Life and General Insurance Industry. We bring forth some of the major applications in the Life Insurance industry in this article. The Life Insurance industry is already using AI in the fields of Client Acquisition&Onboarding, Pre-Approved Sum Assured to clients, Customer Propensity identification, Product Pitch prediction, Persistency Management, Cross-sell/ Up-sell, Risk Management through predictive modelling of bad claims etc. Companies have created dedicated Analytics &Business Insights functions to look at business in a very specific microscopic view. Amrit Singh, Senior Vice President & Head – Strategy, Investor Relations and Analytics for Max Life Insurance says “we have created a dedicated and specialized team “AI works” which is responsible for developing and delivering cutting edge Business First AI solutions. This shows our commitment and focus to ensure Max Life remains an Analytics focused organization. Client Acquisition & Onboarding : AI has the ability to identify the right customer, his preference and his financial capabilities. “We have models that predict customer propensity to buy, recommend the right product as well as the optimum ticket size,” says Francis Rodrigues, who is an SVP, Business Insights, Data Labs&Innovation for HDFCLife Insurance. “The predictive models that we have used at the time of onboarding have helped us to save more than 6 Crores on an annual basis (increased conversion and lower early claims)” continues Francis on the onboarding process. Cross-Sell / Up-Sell& PASA Campaigns : AI can be used to predict customer behaviour from his past dealings with the companies and through various business transactions and stages of life analysis. This assists companies in designing more scientific and data-driven campaigns for cross-sell / up-sell and Pre-Approved Sum Assured (PASA) business. “We use different types of interactions and transactions with customers, payment details, CRM  data, credit bureau ratings etc. and understand the customer sentiments based on which we have created our Aureus SentiMeter”, says Dr. Nilesh Karnik, who is a Data Scientist with Aureus Analytics. Amrit Singh states that they run several PASA campaigns with their Banca partners, using a comprehensive ML predictive model engine, which not only filters most suitable customer profiles with pre-approved sum assured for PASA offers but also recommends the most suitable product. Francis Rodrigues says “We do have cross-sell models that take customer demographics & interactions as inputs and decides the best segment for cross-sell. Our Cross Sellpropensity models have provided a 42% lift in the appointment ratios”. Persistency Management : Usage of AI took a big leap in the Persistency Management process of most Life Insurance Companies in the country. This is clearly visible through the published persistency reports, which shows significant improvement in persistency ratios of various insurers in all the buckets from 13th month to 61st month.  AI tools and techniques have assisted companies in identifying their priorities and the right set of customers through very good predictive models. Francis says that they have a host of models that predict customer’s propensity for renewals and decides the persistence strategy that is to be employed for different segments of customers. Amrit states that their models are built using advanced machine learning algorithms like gradient boosting, support vector machines and deep neural nets. Post issuance we are using host of predictive models to enhance customer servicing, retention efforts and cross-sell / up-sell campaigns. “We use predictive models with artificial intelligence-based customer retention solution, which enables the renewals team to make intelligent decisions by optimizing customer retention effort through focus on high-risk customers. At the heart of our artificial intelligence solution is combination of Deep Learning classification algorithms – FNN (Feed Forward Neural Network) and LSTM (Long Short TermMemory) algorithm and is able to deliver 99% accuracy” says Amrit. He continues by stating that the outcomes of their AI-based customer retention model indicate a substantial reduction in renewal costs without any drop in renewal income. Amrit says “Reaching out to the customers for reminders, only when necessary, has resulted in enhanced customer experience reflected in higher customer satisfaction and better NPS scores”. Assisting Business Processes: Life Insurers have also started using AI in Sales Team Recruitments and Agent Onboarding process.  Also, they have started using various AI tools in improving business quality and revenues through right pitching. Francis says “Apart from customer analytics, we use predictive modelling and machine learning to identify agents / FLS who are at the risk of attrition. This enables us to run proactive interventions to reduce the attrition and retain talent”.Amritreaffirms that even Max Life uses AI techniques for the same and also adds further that “By using AI techniques like Deep Learning and NLP, we have now reached new levels of entire value chain to enable revenue enhancement activities, to reduce operational cost, to reduce friction and enhance customer experience and to acquire superior quality of book. We run smart prospecting campaigns across various digital platforms which have affinity towards financial services to create insurance awareness among the relevant audience. We use algorithm-based targeting and purchase models that allows us to reach out/re-market to our target audience at the most optimized costs. We have tie-ups with leading Web Aggregators and Digital Brokers like Policy Bazaar and Cover Fox to reach out to customers looking for purchasing life insurance online”. Advanced & Scientific Risk Management Risk Management has been the core of all insurance companies and there has always been a focus on this aspect. While Underwriting is the major filter before issuance and the Claims is the final filter before a claim settlement, AI has brought in a new flavor to these filters using a more scientific logic. Now insurers use a trend analysis and past experience using multiple variables and are able to predict probable frauds.  Thanks to Artificial Intelligence. As this matures further, this will be the real future of underwriting in the coming years. At HDFC Life, the risk fraud claims are calculated to screen customers using early claims models at the customer on-boarding stage itself.  “Apart from the early claims models, we have made few in-roads in building in-house artificial intelligence capabilities, which are in various stages of testing and deployment, like Face-Sense and Emolyzer / Sentilyzer,” says Francis.  He further adds that “Face-sense uses computer vision capabilities to identify customers & compare them against their KYC images, acting as an additional layer of risk mitigation while processing payouts at branches, while Emolyzer/Sentilyzer uses Voice and text analysis capabilities developed to identify the emotions or sentiments basis their voice-calls & email communications”. Similarly, the policy issuance process of Max Life also uses a predictive underwriting engine which assesses the 360-degree risk of likelihood of an early claim, lapsation and likelihood of fraud upfront and recommends policies which may require additional verification by an underwriter. Amrit says “We have a two-pronged approach towards reducing frauds – one at the Underwriting stage, as mentioned above and the second, at claims stage, where we are aiding our claims team to identify possible fraudulent claims and thereby focusing investigation only on such claims rather than on all claims. This has enabled us to not only proactively catch risky and fraudulent policies at the issuance stage rather than rejecting them at the claims stage as this comprehensive engine includes models for medical risk, financial risk, Persistency and customer propensity to buy LI policy, but also in becoming industry number one in claims paid ratio”. This is just the beginning.  Lot more to happen in this area.  While Robotics (RPA) is already started in the Banking sector, probably we shall see more of that in the Insurance sector as well. A combination of various tools of AI, including RPA, may play a major role in Financial Underwriting going ahead. We may see more of mathematics and statistics based AI in future which will be used to design various insurance products. Let’s look at more of AI and the confusing interchangeably used term ML in the next edition as well.  To conclude the first part, let’s understand that Machine Learning is a basic logical application of Artificial Intelligence which is a study of various algorithms which automatically improve by learning each experience. So let’s not forget one basic philosophy.  There are limitations to Artificial Intelligence as well.  It is about learning from existing.  If existing is an error, the resultant will be an error. It is just common sense that an Artificial Intelligence or an Artificial Superintelligence is just a hypothetical replica of the thoughts of the person/s involved in the designing of the same. ….. to be continued