Why The Insurance Industry Is Turning To Machine Learning

The industry is often slow to react to technological advances, but could they finally be ahead of the curve


The insurance sector is not traditionally ahead of the curve when it comes to new technologies. In a recent survey by Willis Towers Watson (WLTW), 58% of senior executives in the sector acknowledged that they lag behind other financial services sectors in terms of technology adoption - with digital technology a particular problem.

There are a number of reasons for this. They are an inherently risk-averse group because of the nature of their work. Implementing new technology is always a risk - particularly for insurance, which is besieged by regulations and requires a great deal of transparency - so it would be logical that they would be more cautious. But this is beginning to change, with VC investments in insuretech (technology based solutions for insurance) rising from $130 million in 2011 to $1.7 billion by the end of 2015, and it looks to be on an upward trajectory.

These new technologies are wide in scope, and many are already being deployed. Possibly the most important of these, though, is machine learning and AI.

AI is tailor-made for the insurance industry because, as Adam Devine notes on VentureBeat, ‘adoption of AI and automation will be highest in regulated industries and those that must process thousands of transactions and customer requests daily.’ Japanese insurance company Fukoku Mutual Life Insurance, for one, has already laid off 34 employees and replaced them with an AI system that can calculate payouts to policyholders.

The system Fukoku Mutual Life Insurance uses is based on IBM’s Watson Explorer, which utilizes ’cognitive technology that replicates human thinking to ‘analyze and interpret all of your data, including unstructured text, images, audio and video.’ According to local newspaper Mainichi Shimbun, the technology allows the company to read tens of thousands of medical certificates and factor in the length of hospital stays, medical histories and any surgical procedures before calculating payouts.

Fukoku Mutual Life Insurance believes that they will be able to increase productivity 30% and make 140m yen worth of savings per year, all for only around 15m yen a year.

Machine learning and AI has a number of other applications in insurance. The most obvious areas it can be used include claims processing, underwriting, fraud detection, and customer service. They can also look to add wider benefit to the business by analysing market dynamics, competitor activities and, customer trends, detecting patterns in the data to garner insights with unprecedented granularity and at a speed humans are simply incapable of.

Fraud Detection

There are 350 cases of insurance fraud worth £3.6 million uncovered every day in the UK, while in the US fraudulent claims across all lines of insurance total some $80 billion a year. This number is only going up, and the overall cost of investigation rising with it. Machine learning is able to auto-validate policies by ensuring that key facts from the claim marry to the policy and should be paid. Once validated, this information can then automatically be fed into the downstream payment system and money sent in a matter of minutes without any human involvement.


Data is now being mined from a variety of sources that can help insurers build a fuller picture of their customers. Machine learning algorithms can analyze this wealth of information quickly and accurately, without being tainted by human bias, and help to offer more accurate prices. In health insurance, for example, data from wearable devices such as Fitbit can track a customer’s activity, while analysis of their social media may give a more accurate idea of somebody’s lifestyle choices than they are willing to share. This will likely punish those who are unhealthier than they say, but it will also reward those who live healthier lifestyles.

Customer Service

According to a 2014 Capgemini survey, a mere 29% of customers are satisfied with their insurance providers’ services, globally. AI can help improve customer service in several ways. Firstly, by driving chatbots that customers can engage with during purchase and claims processes. The time it could save insurance agents in not having to answer routine questions or carry out basic admin tasks will also free them up to provide a better service.

Another immediate benefit is in the ability to better monitor and understand interactions between customers and sales agents. They can analyze recordings to detect patterns in how customers respond to certain calls so that improvements can be made. These recordings will also help to improve controls over mis-selling of products, which greatly aids compliance. They can do likewise with written documents. Captricity, for one, converts unstructured data from handwritten and faxed documents into structured data using AI that this can be mined to discover insights about customer.


In the WLTW survey, 42% of senior-level executives in the insurance industry cited the complex regulatory requirements as the largest barrier to adoption of digital solutions. Keeping an eye on regulatory changes is a pressing concern for insurers anyway, and regulatory bodies are usually well behind when it comes to embracing technological innovations so there is always a chance that, once adopted, changes will need to be made to keep pace.

There is also issues around what happens to the humans currently employed in insurance. A significant portion of any company’s costs is staff, and insurance is no different. Machine learning will replace many menial tasks at first, and more complex tasks later, potentially helping to save insurers millions of dollars. As we have already seen with Fukoku Mutual Life Insurance, this is an extremely attractive option and many are likely to take it. Whether staff agree that it is a good thing is another matter, and senior management will have to be transparent or risk causing serious issues.

Ultimately, however, machine learning brings with it many benefits, and insurers will ultimately have to embrace it, as every industry will, in order to retain their competitive edge. However, they will have to exercise caution and make sure they invest appropriately, while also retaining key knowledge and skills among their staff and meticulously tracking any new regulations in case anything goes wrong.

Big data hype small

Read next:

Is Big Data Still Overhyped?