AI enabled transformation of insurance
The computer, a dog and an actuary
Have you heard the one about the insurance company of the future? It consists of three things; a computer, a dog and an actuary. The computer runs the AI enabled insurance company. The actuary feeds the dog. And the dog is there to bite the actuary if they ever touch the computer!
It’s not a joke I’d recommend for a first date unless you’re specifically attracted to data scientists. But underlying the attempt at humour is a serious observation. That the insurance company of the future will (predominantly) be an automated AI operation that is supervised and/or instructed by a human.
But don’t panic just yet, in reality, there is still a way to go before this robot vision of the future is realised. And even when it does, it will be a hybrid model where the functions of insurance business are split between the tech and workforce. It was a subject I covered recently in this conversation piece with Steve Mendel of Bought By Many, Geoff Keast of Montoux and Dr Mark Farrell about the digital actuary.
Digital or die for insurers
Whilst there are many predictions for the future, the one thing that we do know for certain is that the world has changed. The next, and many of the current, generation of insurance customers live their lives digitally now. They don’t buy records or CDs when you can stream music on the go from Spotify. They no longer stand in the pouring rain hoping for a taxi to pass their way when they can Uber a ride in precisely 3 mins. And when they want to get away, they go and Airbnb in somebody else’s home. Digital is here to stay and insurers have no choice but to adapt.
This is the double-edged sword of opportunity and threat that insurers face today. Staying analogue is a sure-fire path to obsolescence. Going digital opens up the established fundamentals of the way the industry works.
AI-driven pricing makes it easier to sell insurance
From Cuvva to Trov to Lemonade, the InsurTech start-ups have shown that it is entirely possible to automate risk rating and premium pricing, and do it in a way that is quick, efficient, effective, convenient, as well as 100% digital.
Take Lemonade, for example, an InsurTech favourite of mine and the subject of several posts. As Daniel Schreiber, their CEO and founder once said, “we have bots instead of brokers, and an app instead of paperwork.”. Which is why they can write personalised cover in less than 2 minutes (using AI bot Maya). When it comes to claims, for the majority of cases they settle just as fast using Jim, their claims AI bot.
AI surveillance will improve the claims experience
The automated Lemonade solution is a great example of the future for the handling of some claims, but not all. There are many cases where customers will want to talk to another human being (not an AI-powered bot).
And for these digital customers, insurers are going to have to make the claims experience so much better than it is today. I like the direction of travel of Dave Stubbs, the co-founder and inspiration behind InsurTech claims platform RightIndem when he describes “making a claim should be like having a chat with a friend down the pub”.
The concept is really simple. Digital customers want to tell their story in their own words, not be constrained by form filling (whether that’s online or over the phone). In an AI enabled claims operation, the customer can relay their situation in their own words and the agent can (just) listen. Because the surveillance algorithms are listening too!
The algorithms are listening to every word the customer says, filtering out the superfluous noise and collecting only the information the claims agent needs (just as Alexa is doing today).
The AI populates the claims workflow with the information it ‘hears’ from the conversation. And then prompts the agent with clarification questions to ask the customer to complete the gaps. The AI also checks external data sources for corroborating evidence, such as the weather report for last Tuesday at 8.30am in Northampton to confirm that it was raining heavily when you skidded off the road into a tree!
This is a powerful hybrid operation between man and machine. As the human agent listens empathetically to the customer’s story, the AI’s algorithms identifies the gaps, or possible conflicts and inconsistencies in the story. As a result, the AI driven claims process directs the agent to steer the conversation with the customer in a specific direction.
The end result is that the customer (should) feel like they are have had a real conversation with their concerned, helpful and interested insurer rather than an interrogation. And, of course, the insurer gets the information they need.
Augmenting reality with AI and machine learning
Data has always been at the heart of the insurance industry when it comes to understanding and assessing risk. Now, AI enabled surveillance offers insurers a perspective on risk at a granular level.
Take health insurance, typically a single point in time, blunt instrument assessment of a customer’s state of health. The cover may run for many years without ever reflecting any changes in the customer’s risk profile. But with technology, this will all change. Whether that is detecting early signs of Parkinson’s disease from subtle changes in the way you move around a keyboard on your computer to your smile on a selfie indicating you are suffering from a mental health condition.
The AI that processes this data and identifies the impending early stages of a health condition may ‘know’ before you do that you have a condition. This is the kind of highly personalised perspective that Google and Facebook have the means to know about those that use their platforms (which is a lot of us!).
For the underwriter, this knowledge would be invaluable if they had it and were able to act upon it. Although it won’t be so good for the customers who suffer adverse selection when insurers hike their premiums or refuse cover. Thankfully, the regulator and the data science community have this issue in their sights. For more on the trend towards the personalisation of insurance, read my recent article with insurance ethics guru, Duncan Minty.
The automation of fraud detection using AI
One area of the insurance value chain that has benefitted from the application of new technologies, and especially when it comes to AI, is fraud detection and prevention. According to the FBI, fraud costs the US insurance sector in the order of $40bn.
One of my InsurTech favourites in this space is Shift Technology. What they’ve built is a decision-making platform speciﬁcally designed for insurance fraud handlers to augment their capacity to detect fraudulent behaviours. This is a hybrid man/machine solution that does not replace the human fraud handler but uses the might and power of AI to replicate their deductive reasoning and apply it at large scale.
Shift’s AI driven fraud detection solution is named Force and it is already deployed at more than 70 insurers around the world. The solution assesses copious volumes of big data to generate a fraud score for every claim, achieving a 75% average hit rate.
What do the experts think about the impact of AI on insurance?
To find out, join The Digital Insurer on April 30th as we host the latest in our series of informative webinars on this very subject. It’s free to attend and you can find all the details here.
Related articles of interest
The author, Rick Huckstep, is an InsurTech thought leader, advisor and speaker. He is the Chairman of The Digital Insurer.
Disclaimer: Rick Huckstep is a investor in RightIndem.