The insurance industry is probably not the first area of human endeavor one thinks about when somebody mentions words like “innovation,” “artificial intelligence,” and “high tech”. However, once this industry embraces the change of AI, you will see this comprehensive integration is not less competitive than other industries.
Previously, we have discussed about AI in Healthcare, AI in Supply Chain which showed an enormous opportunity of AI in the 21st century. Insurance sector, is a promising sector with endless opportunities. Insurance is, in very simple terms, an industry built around risks. Insurance companies greatly depend on their ability to predict risks from one person, company, or organization. The more information they have about them and the more accurate this information is, the more likely they are to make a correct prediction, either saving themselves money or earning extra revenue. In recent years, technology-enabled innovation in insurance is highly welcomed, as in 2016, USD 1.7B across 173 deals was funded on these new insurance technologies (Deloitte, 2017).
The emergence of AI technologies means that insurance companies should scramble for the ways of implementing them in their work to get the much-needed edge over the competition. But how exactly do these innovations change the industry? Let’s have a look at some major areas that Artificial Intelligence embarks to Insurance sector.
Behavior-based premium pricing
One of the most obvious examples of insurance industry technology that completely changes the way things are done are telematics and wearable sensors collecting information about customers. The resulting avalanche of new data created by these devices will allow carriers to understand their clients more deeply, resulting in new product categories, more personalized pricing, and increasingly real-time service delivery. For example, a wearable that is connected to an actuarial database could calculate a consumer’s personal risk score based on daily activities as well as the probability and severity of potential events.
Currently, financial models are mostly built based on statistical samplings of past performance — that is, companies study the client’s record and build their predictions upon it. This new approach allows for real-time, current information to be received and used. No longer will careful drivers have to pay extra for the less careful ones because the offers can be individualized for each and every customer. The benefits of pricing client’s premium based on behavior are numerous for both insurers and insureds:
- Encouraging better driving habit
- Lowering claims costs for insurers
- Changing carrier to customer relationships from reactive to proactive
The insurance industry already actively uses chatbots — they help build up the initial communication with the customer without having to resort to human employees whose efforts may be better applied elsewhere. This approach allows for moving the entire interaction between the company and the client online, dramatically decreasing operational costs and thus lowering the price of premiums. And any company that only works with its customers online has to rely on machine learning to prevent fraud and guarantee that every customer gets individualized experience.
Utilizing AI and machine learning, chatbots can interact with customers seamlessly, saving everyone within an organization time – and ultimately saving insurance companies money. A bot can walk a customer through a policy application or claims process, reserving human intervention for more complex cases.
Faster claims settlement
Two of the most important factors defining the efficiency of an insurance business is how fast it manages to settle claims, and how successfully it does it. Introduction of AI dramatically boosts both of these factors. Insurance companies have massive amounts of data. On its own, data doesn’t provide much benefit. AI can process data quickly, helping insurers to automate and accelerate the claims process which is faster than the best human supervisor can ever hope to achieve, as it rarely takes the less than a few weeks, resolving in better insights.
Decreased fraud occurrence
It is physically impossible for human insurers to gather and process all the information about policyholders that can be an indication of fraud. Companies that rely on AI solutions are capable of processing virtually unlimited amounts of such information, which means that claims are settled not just faster than it is done traditionally, but also with a much lower percentage of fraud. Additional use of machine learning for fraud detection also means that AI learns to improve their results over time, getting the ability to notice the telltale signs of fraud more efficiently as they encounter its new and new instances.
Statistics show that up to 10% of claims costs are related to fraud. Using predictive analytics, AI can identify more fraud than a human adjuster.
As AI becomes more deeply integrated in the industry, carriers must position themselves to respond to the changing business landscape. Insurance executives must understand the factors that will contribute to this change and how AI will reshape claims, distribution, and underwriting and pricing. With this understanding, they can start to build the skills and talent, embrace the emerging technologies, and create the culture and perspective needed to be successful players in the insurance industry of the future.
Tell us about your business needs!
FPT highlight case studies in Insurance Sector
The process of insurance includes several known steps (Figure 1), starting from Product Development to Marketing & Sales, to Underwriting & Risk Management; following by Customer Servicing, Claim Management, Financial Assets and Operation. FPT’s ability in Insurance sector can also be shown in detailed cases below.
Figure 1: FPT’s ability in Insurance sector
Currently, the underwriting process requires customer to fill in forms. After that, the evaluation will be done by the system with fixed rules. If a profile is rejected, the company may not know the specific reasons. Not to mention, the fixed system has low flexibility and cannot evaluate new cases with new pattern effectively. New core AI model has already been developed in the company but is not optimized yet.
FPT proposed to send a team of one solution architect and two ML engineer to work closely with the customer to deploy the system. There are two new components that FPT need to develop: a REST API that wrap the core AI model and a Traceability Web to track the predictions of the core AI model. Our team’s deliverables are:
- All source code, included unit test
- Manual guide documentation for Traceability web
- REST API documentation
The model is now capable of processing with high accuracy and at a rapid rate: it will take 1 second to process 1 profile and only 2.5 seconds to assess 10 profiles.
A company is interested in implementing a straight through process (STP) system for health insurance claim processing. The system needs fraud detection engine for health insurance claims. FPT proposes to perform a PoC (proof of concept) to implement a machine learning model for fraud detection engine using sample data from the company. The purpose of this PoC is to assess the feasibility of automatic fraud detection engine using Machine Learning. If it is successful, the PoC will be followed with a limited pilot, then next a production deployment. The company improves processing speed, accuracy with lower cost, drives to higher customer’s satisfaction. The process for health insurance claim can implemented in a massive scale.
Telematics Insurance is a near real-time monitoring driver system used to analyze driver’s behavior, including emergency alerting, tracking vehicles. The solution provide by Telematics Insurance can be applied to insurance company, emergency company, car rental or any company want to manage drivers.
This system is one of insurance services which payment is based on driver’s risk score. Using AI, FPT calculates the risk score based on the frequency of using brake, driving speed, ... The system keeps information of all registered driver and make reports over it. The reports concern personal information, driving information, event trigger and evaluate the risk of the driver.
With this information, company can manage their driver and analyze driver’s risk level based on their driving behavior. The simulator in administration level can simulate a scenario to track a driver. By using brake, shock, B-call, E-call button, all the events are saved and send to monitoring interface. When an emergency call is triggered, administrator receives the location, images and an email concerning the emergency. The system provides the ability to retrace the driver route until the emergency is reported. Administrator can use this to help the driver in case of emergency.
The company sells car insurance to customers through two channels: direct contact or through calls. Current procedure for handling an accident is time-consuming and reduces customer’s satisfaction. Besides, the insurance accounting department also takes time to handle payment processing. This can take more than 1 week to complete one case. To help our customer, FPT does:
- Increase communication channel with customers: by creating chatbot module, customers can record accident scene then receive response from insurance company promptly.
- Use AI to analyze data through data lake and judge the extent of damage and issue compensation amount, the customer will receive compensation upon agreeing to the amount.
- Gather customer opinions through social networks thereby obtaining results of the accuracy of AI when making reasonable compensation.
Using AI brings great benefits for the company. The waiting time is reduced considerably: from one week to just within 30 minutes. Customer churn rate is expected to reduce from 3.5% to 1.5%, which exceeds the company’s expectation.
OCR in Insurance
The customer is one of the largest insurance companies in Asia which has been helping people to protect their possessions, themselves and their families, and look after their money for about 100 years.
The paperwork process for customer’s insurance claims has many issues since it is labor-intensive. Once customer submit papers such as hospital admission, hospital expense bill, … which are not the same forms and contain complicated information in different hospitals, the employees have to check whether the files are the right forms and the information in each file is correct. This is time consuming and not optimized labor resource’s capacity, furthermore, causes mistakes when checking and assessing papers.
FPT applies auto OCR to read forms automatically, then develops an engine that can export information from form’s fields to database system and classify medical documents. If the uploaded file is wrong, it'll notice and reject immediately. The information from a correct file will be automatically extracted for the employee to check easily.
Applying OCR to solve issues of customers' insurance claims helps reduce workload for employees and increases accuracy significantly, which a human employee cannot complete within seconds.
In Insurance Industry, FPT has embarked on implementing many outstanding AI projects with are the current global trends, from behavior-based premium pricing, fraud detection, insurance claims settlement. Stay tuned for more new activities from FPT in the field of Insurance in the future!