Sample Solution for AI course

Use of AI in Insurance

ABC Inc is an insurance company in the southern province of Lebaly. The company provides insurance solutions to employed people and has a variety of schemes to suit various requirements. [a para on the company relating to insurance]

As their customers are increasing, one problem they are facing is in helping the customer choose the right insurance option. There are only a handful of agents qualified to help in this problem, and they are always busy. As a consequence, adequate attention is not being given to people and often incorrect choices are made. Long waiting time for getting attention of an agent is another problem. This is turning off many potential clients.

Simple automation will not work here, due to the nature of the problem. It is not easy to map available schemes to requirements as a simple table of policy amount or a small set of such parameters. The schemes vary in the kind of returns one would get on maturity and also after a specified lock in period. One can choose a scheme with a monthly return after 10 years with a larger lumpsum at the end of 25 years, for example. This would depend on the user's nature of need. Certain schemes are not available for people with certain profiles - like salaried people, business men, daily income people, and so on. The returns are higher if the person is willing to take more risk. A richer framework than conventional decision tables are needed to solve this problem.

[similarly do one more problem and analysis]

One AI technique that one can use here is expert systems. Expert systems behave like a human expert solving problems generally in very narrow areas like disease diagnosis, equipment troubleshooting, etc. They capture the key knowledge in these areas in the form of if-then rules and use them to make inferences. Mechanisms to support uncertainty handling (yellow eyes are likely to indicate jaundice), to offer explanations (how did the system arrive at this decision), etc are often seen in expert systems. [a little more explanation]

Expert systems are an appropriate technology here. The domain is very specialised and fairly narrow - the task is to seek relevant information from the user and recommend one or more schemes. This is a problem like diagnosing a disease or recommending a type of plastic - areas where expert systems have been used in the past. The existing agents can be consulted to acquire the knowledge, and they can be represented as rules. "If the person has high-risk-taking nature, then recommend stock-based options". […a little more explanation…]

[do one more technique]

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License