AI is going to disrupt almost all verticles in coming years. In health insurance industry there are many ways AI can change the way things work. If we take end user into account then AI based Insurance Agents are already being testd upon. These agents have intelligence of various products that a company has to offer and can give judgement based on persons lifestyle.
Insurance companies are also working on smart products that can be highly customized by understanding user need. Due to this there will be no fixed product names that company will have to offer but the product would be fully customized based on person's need and current family situations.
Predictive analytics is commonly used in cost projections while selectibf a policy baed on family health history, financial capability, location etc. Deep learning based models are also being developed that will help in making better choices when it comes to policy selection.
In a nutshell we will see a new breed of insurance products in future that are more consumer friendly and cost effective.

There has been a very interesting shift in insurance because of AI. Think about autonomous cars: what insurance people can get in case of an accident? There are rumors that autonomous car will have a black box, similar to what exists on planes, where everything is stored through the sensors that exist inside the car (people counting, sentiment analysis, drowsiness detecting, eyes tracking, etc).
Because of such new paradigms, insurance companies need to be prepared to embrace these trillion dollar market in the most effective way, which will translate in two scenarios: going bankrupt because they didn't do a proper risk evaluation, or succeed because they planned these new market entries.

I feel there are two parts for AI to implement in Insurance Industry:
1) Preparing for AI based system for claims
-Digitization of Insurance industry: The claim input framework, processes to adjust, award and decline the claim.
-Looking at your architecture and building AI-driven prediction models this may include many technologies and products but at the base level you have to look at -
Running a Hypothesis:

• have a solution where you can run an hypothesis (Hadoop HDFS with Spark Machine Learning Engine or similar other products like HPE Vertica).
• Integration of the common database for your core engines like SAP, Oracle or SQL database (structure data) to import to Hadoop Distributed File System
• Creating a central processes with at the edge sensors (IoT) and Augmented reality to enhance the processes for claim adjusters and other operational people.

Testing the Hypothesis

• Once you test the hypothesis, go back and look at what is working and what is not and what are the reasons for it.
• Re adjust the model to have better results.
• Validate databases and HDFS, Sandbox development working with Insurance SMEs and Claim adjuster and IT staff.
• Come to conceses

Create Raod Map

• Implement the roadmap and have the ability to adjust.

2) Smart algorithms for identifying what is working and what is not, new trends, and finding incorrect claims. AI approaches with algorithms to identify only those claims for which the likely-hood of successful is high and adjusting the price accrodingly and then find the way so to lower the likely-hood and probability by build indicators to help and adjust the claim to lower the incident of that event is the way where the industry will be.

It will become predictive, suggestive and informative in nature, off-course the pricing will go lower for clients but the base will increase and insurance companies will be able to innovate to come up with new products and service from this intelligence which will be gathered and analyzed as an ecosystem.