Applying AI in Insurance

AI in insurance

 

5 MINUTE READ       

 

By Daniel Thornton

 

According to Wikipedia the term ‘artificial intelligence’ refers to a “machine that mimics cognitive functions that humans associate with other human minds”. At present, some of the things we associate with other minds, such as learning and problem solving can be done by computers.  Although they do not process information in the same way we do.

 

According to PwC, AI now represents the biggest commercial opportunity in today’s economy. Their recent report showed that:

 

“The main contributor to the UK’s economic gains between 2017 and 2030 will come from consumer product enhancements stimulating consumer demand (8.4%). This is because AI will drive a greater choice of products, with increased personalisation and make those products more affordable over time.”

 

This change in the market is bound to impact the insurance industry too. Given this, many forward-thinking insurance companies now have AI on their drawing boards. In this post we’ll explore 3 areas of insurance which are now ripe for the application of AI technology.

 

Chatbots

 

In the past, many insurers have been accused of taking a distant and impersonal stance towards customers. The reality of the situation is that most insurers are simply unable to handle the huge volumes of customer queries they receive about their products and services every day. This is where chatbots can make a real difference.

 

Chatbots are computer programs which can conduct a conversation via text. These programs are designed to convincingly simulate how a real human would behave.

 

A chatbot can be used to engage the customers by providing accurate and relevant information. As they are machines, using them means doing away with human bias, error and fatigue. What’s more, they can attend to multiple customers at once, at a time and place convenient to the customer.

 

This technology has already been deployed in the UK. In September this year, Co-op Insurance made the announcement that they were the first company to launch a car insurance chatbot. The bot, which can be accessed through Facebook Messenger, asks customers 4 simple questions and can return a policy estimate within 30 seconds.

 

 

In Underwriting

 

AI in insurance

 

AI is used to build Deep QA Systems. These rely on complex question and answering techniques that help underwriters look for appropriate indicators of risk. This helps the insurance companies assess their risk better, improving their profitability. For instance, if the AI-engine finds out that a certain customer had filed for insurance repeatedly in the recent past, this will be flagged up for further investigation.

 

Soft robotics use process mining techniques to automate operations and improve efficiency. For instance, if an AI engine finds out a customer has missed previous premium payments, the AI engine itself can set up text or email reminders that help both the customer and the company.

 

Machine Learning involves using decision tree analysis and deep learning to develop predictive models of risk. For instance, it could build a model which determines the likelihood of insurance misuse based on the traits of previous customers and then apply that model when processing new customers.

 

However, companies need to be careful about the kinds of information they are using to construct their model. Last year, Facebook pulled the plug on a machine learning project devised by Admiral Car Insurance. The plan was to ask drivers for permission to evaluate the language used in their Facebook posts in order to try and get a potential discount on their premiums.

 

Automation in Claims Processing

 

Process mining techniques can also be used to identify bottlenecks and improve efficiencies with standard claims processes.  For instance, repetitive and mundane tasks of the process can be automated saving huge amounts of time and effort. Furthermore, since automation does away with human bias and error, it also improves compliance to existing processes.

 

Machine learning can use the data a company already has and apply it to determine the costs of covering a new claim. Such techniques are already being used to categorise the severity of damage to vehicles after a traffic accident.

 

 

What does the future look like?

 

According to the PwC report: “AI’s initial impact primarily relates to improving efficiencies and processes. Over time its impact will be more profound; it will identify, assess and underwrite emerging risks and identify new revenue sources.”

 

 

To browse our latest big data and data science jobs, click here

Or apply now for:

Data Scientist – Machine Learning, PySpark

Big Data Engineer 

Big Data Architect