Apr 15, 2019
The insurance industry has been slowest to embrace big data and artificial intelligence (AI). Today, there is a tremendous amount of data available on everyone seeking insurance. Reviews on Yelp and TripAdvisor will tell you if a restaurant serves ‘generous’ drinks and whether guests felt safe in one of your insured hotels. This data coupled with more traditional underwriting data of course makes for better insurance solutions. So why the reluctance from within the industry?
Where underwriters traditionally have relied on data about demographic, geography, gender and generalized habits and behavior for an overall group, today they have access to information that goes much deeper. This is big data. The availability of data from telematics, wearables, Facebook, etc. is allowing insurers to customize products and ‘underwrite for you’, not just ‘someone like you’. Amazingly, handful of insurers have actually got down to implementing data sciences technology. Here’s the thing about big data.
Big data is messy and that’s an understatement
There are several components to big data. One part of big data is structured and maintained in neat and easy to find places (name, address, years in business, payroll of policyholders) and easy to process. But the other part is unstructured and complex to synthesize, the information gleaned from human interactions. In emails, chats, reviews, data is not neatly labeled, but these exchanges carry a wealth of information about our lives. This information comes fluidly, in sentences and natural language. It could be a chat about a contractor having to rework electrical connections, increased break-ins in a warehouse district, or employee feedback on Glassdoor. This is invaluable information. But it is also what makes big data messy and often times excluded from being taken into account.
So how to sift through this mess? Here’s where Artificial Intelligence comes in and extracts insights from the mountain of information.
Big data leads you nowhere without AI
With everyone on the internet all the time, imagine the amount of data being generated. It’s overwhelming. Thankfully, AI is more than equal to the task of processing it, gleaning meaningful insights from it that lead to better decision making. Consider you receive a new warehouse risk from one of your agents. Historical loss history is average and nothing seems out of the ordinary. However, utilizing big data, you can tell that this warehouse risk has many negative comments on Glassdoor. Moreover, you see that they are related to a new manager and his reluctance to provide adequate training on equipment and a disregard for safety rules. In this instance, future performance will likely be different from past performance. Artificial Intelligence, applied to big data, will help draw out these correlations in real time and help the underwriter make more informed decisions. Given the amount of work that crosses an underwriter’s desk in a day, it would be impossible without the right technology and AI to utilize the additional data.
AI is designed to put two and two together for you. It can sniff out hidden links between data points and decipher patterns within a fraction of a second. Therefore, saving an underwriter’s time and effort on some of the crucial, but menial tasks of searching and aggregating information.
The best time to get in on the data game? Yesterday.
The additional information brought by Facebook, Twitter, Glassdoor, Yelp, etc. have already transformed underwriting. And there’s more on the way in the form of information coming from smart devices, telematics, wearables and Internet of Things. Structured data might once have been our comfort zone. Now it’s time to face the future.
AI is not your enemy. In fact, it could be your BFF.
Understandably, the mention of AI brings in a feeling of dread in every industry. But an underwriter’s judgment is still the key. We are still some time away from AI that even approaches human intelligence. What you see in sci-fi movies like Ex Machina is just that, science fiction. But right now, technology can utilize big data and AI to carry out the kind of tasks that underwriters spend almost 70% of their time doing: background work and low-level tasks like searching, aggregating, selecting and reviewing information. This type of work is time consuming and inconsistently applied depending on the individual.
We often speak of underwriting as an art. And it is, when it comes to using human judgment and instinct. But why not use science to brighten the colors?