Artificial Intelligence in the World of Insurance
In the blog ‘Driving Opportunities for Insurers through Artificial Intelligence’, Subramani Rajamani has expressed his views about the rise of intelligent systems through consolidation of digital technologies such as Internet of things (IoT), social media, analytics, data storage and Artificial Intelligence (AI) technologies. He also detailed the building blocks of AI such as text analytics and Natural Language Processing (NLP), audio/image/video processing, Robotic Process Automation (RPA) and Machine Learning techniques. In continuation of his views, let us discuss a few interesting AI models and ideas that we envisage to be potentially explored by insurers in the near future to address the needs of customers and internal stakeholders.
It is quite common to notice that few diseases are being more prevalent in some regions and at certain periods of time or seasons. An eventuality due to diseases can vary widely based on geography and weather conditions. For example, consider Dengue. The outbreak patterns of dengue can be analyzed basis geographical areas and periods (post monsoon). Geographic specific products can be designed to accommodate the most prominent causes of illness based on such analysis and priced accordingly. AI programs trained to monitor, analyze, evaluate and learn from such information can prove helpful in providing location and weather-based coverages and prices.
New Business Done Differently
There is nothing better than AI enabled advice-driven buying process making customer’s life easy.
Insurers can integrate the weather forecast models to their lead generation system to recommend coverages to customers during their buying journey. Similarly, in case of life insurance, online distribution tools can leverage data from integrated financial information sources to evaluate the customer’s appetite for risk, based on his other investment portfolio and suggest products and coverages accordingly.
Based on the information gathered about a customer, personalized pricing is also possible. Let us consider the personal property for instance. In a flood-prone zone, the personal property belonging to a customer residing in the ground floor is much more at risk compared to one residing in the first floor. Hence, two customers living in the same building could be charged differently for the same product.
Risk Mitigation Re-imagined
How about customers not having to face a catastrophe in the first place and insurers not having to pay any claim? Sensors installed can help identify weather patterns in a particular region along with forecast models to predict perils. This can bring together the geography-based information required for insurers to personalize and establish risk mitigation, loss minimization and evacuation strategies. For example, in the event of forecasting a flood or hurricane, insurers can proactively arrange for people and personal property evacuations for their customers.
The New Age Underwriting
As part of underwriting, an insurer performs a product suitability check for an applicant based on various parameters to assess the risk and to arrive at the underwriting decision. The applicant discloses the requisite information in the application form and the policy issuance and the contract abide by the principle of “Uberrima fides”. In case the applicant had unknowingly missed providing a particular information, it results in a loss for the customer. To improve customer experience, insurers can perform the suitability check through information aggregated from third-party sources based on National ID (Social Security Number – SSN / Unique Identification Number - UIN) of the applicant. For example, World-Check is a database used to identify politically exposed persons and high-risk individuals. This source of information can be connected with insurance underwriting systems to assess the risk related to a person.
In medical insurance, a medical profile can be created for each customer and can be updated regularly through feeds from various health care centers. This can be consumed by an insurance company to evaluate and analyze the risk of an individual based on his visit history to hospitals and clinics and decide underwriting outcome and personalize pricing. Similarly, social media can be leveraged to identify the occupation and hobbies of customers which will help in risk assessment. Frequent travel and adventure trips can be used as inputs to pricing medical, travel and term insurance products. Facial recognition can be leveraged to identify the age of a person or can be analyzed for suspected health problems to help cross-check the disclosed information.
Thus, irrespective of the customer withholding a fact either knowingly or unknowingly, to eliminate moral hazard in underwriting, AI on integrated sources of information can serve as an effective tool to validate data. With blockchain and telematics in emergence, these days are not very far.
The Trending PAY model
The question is ‘Why should I pay the same premium as that of a person who’s traveling 7 hours a day prominently covering accident-prone zones, when I hardly use my vehicle and probably cover few kilometers of safe distances a day?’ With the rise of intelligent questions, it is time to get prepared with intelligent answers.
Traditionally insurers offer a discounted premium or a no claim bonus based on history rather than the real-time behavior of the insured. Today, with the emergence of AI and connected devices, the insured’s behavior can be instantly captured to optimize and personalize pricing. Here are a few examples; usage based pricing is trending among auto insurers. The PAYD or the Pay-as-You-Drive model enables calculation of premium dynamically, based on the amount driven and the nature of the ride. This could be the odometer reading or the miles driven, the speed, application of brakes and accelerator, use of mobile phone while driving, time taken to travel and so on. This data can help insurers assess the risk on an individual basis and change the insurance cost dynamically as per the changing risk. Though we find this more prevalent in the auto segment, the model does have the potential to be embraced by other Line of Businesses (LOBs) such as travel and property through Pay-by-Ride and Pay-as-you-Stay techniques, thanks to carpooling and home sharing services such as Uber and Airbnb.
Good Bye Hassle in Claims
Imagine a claims process without any human intervention. Welcome to possible! AI image processing to automate claims process is the next big thing in the auto LOB. AI trained Image recognition technology to assess the accurate cost of damage due to accidents based on user-submitted photos can help digitalize the claims journey.
Check out this video demonstrating Mindtree’s conversational solution for First Notice of Loss (FNOL) that helps the customer to notify an auto claim through a social media messenger.
Successful Business by the End of the Day
AI tools can be used to track and identify revenue patterns for an insurance company to help better manage revenue and expenses. For example, there could be lean months during a year such as Christmas season when the premium collection could be considerably lesser than other times. A picture on this forecast can help businesses lay down business budgets, product profitability, break even and other parameters to assist the financial and actuarial teams to estimate accurate profitability and viability of a product. AI can learn from the past to better estimate the future and help take corrective actions.
Let us deep dive into more possible areas where AI can be leveraged in insurance value chain in the upcoming blogs. Do you think AI will play a significant role in the insurance industry soon? Do share your thoughts with us.