AI use case in insurance

The Future of Insurance

Foretelling the Future

Artificial Intelligence (AI) can use predictive analysis to identify specific areas of risk, as they apply to each client, a particular geographical location, vocation or avocation, thus making insurance more relevant. That ability allows insurers to offer products that make sense to cover those risks while providing more appropriate pricing.
The problem for the immediate future is learning how to retain new, younger customers; they are the most challenging market segment to hang on to because they want convenience, agility, and personalization—things the industry hasn’t been known for in the past. They expect online service (or through their mobile devices), and presented in a predictive manner so that they don’t have to figure out their own needs.
Just as important to them, they want to be connected to emergency services automatically when necessary. The IoT (Internet of Things) will allow the fire department to be alerted upon need, but a living room smoke detector might not awaken people in distant bedrooms. It should also signal the AI automatically to alert the customers at 3 am to wake up and get out of their house. A car accident should have emergency services dispatched, and communications attempted with the driver and passengers.

Automation of Ordinary Functions

AI can run the backend of the insurance business without people. AI is proving to be good at assessing complex underwriting faster than humans can do it. In the case of aggregated claims, such as hurricanes, hail, snowstorms, forest fires, mudslides, or other natural disasters, claims can be assessed faster and more safely by sending AI drones to survey damage areas. They would take particular note of customer addresses by GPS, but likely be able to contribute overall information for site management, emergency workers, rescue, and so on.

Improving the Customer Experience

People don’t like calling their insurance company about a claim. They fear rate-reprisals, Red Tape, and future insurance refusal. Adopting newly available technologies will offer a facelift for this often “antiquated” industry.
People seldom feel happy after making an insurance claim. It took too long; they didn’t get what they perceived as “full value,” or somehow this insurance they have had for 15 years with sewer backup protection had mysteriously acquired a $1,000 deductible that wasn’t there when they first obtained the policy.
Don’t believe it? Ask yourself how many times a claim sat, unprocessed for weeks, while an agent is on vacation. In truth, the Red Tape phenomenon is the most significant impediment to customer satisfaction.
It annoys the employees who are at their desks with nothing to do. If it wasn’t forbidden by antediluvian rules, they could handle the claims of the employee from the next cubical, who is in Bermuda until the end of the month.
It annoys the customer who has a builder threatening to sue them for non-payment after rebuilding the (insured) back porch crushed by a falling tree. Especially when the payment has already been approved, but the agent has yet to sign off on it.
There are two kinds of claims processing: “Touched” and “Approbation.” The approbation claim, as shown above, can freeze until the right person is available to pass it on to the next person in the chain. The touched claim can be handled by anyone in a department but requires so many transactions to finish the approval process that it, too, can take weeks to work its way through the system. It’s time for the “Touchless” system…


AI managed claims to receive and record damage from the assessor, on a typical form, and then audit the claim, and communicate with the customer; and it all occurs with minimal human interaction. Not only is the speed of processing claims immensely enhanced, but the costs are comparatively small.
Also, you should bear in mind, that AIs can already assess automobile damage claims entirely independently by the photos a person uploads. They are compared to pristine images of that model in a database.
The AI would identify which particular shots and angles are required for the claim by numbers on an unfolded version of the claimant’s model of car. The customer can include any explanatory shots, as long as they are numbered according to the instructions. This provides greater consistency in payouts.

New Abilities

AIs that are explicitly programmed to identify structural damage could incorporate the knowledge of engineers, architects, builders, roofers, landscapers, plumbers, electricians, drainage experts, and any other required professions. Once thus equipped, these AIs are known as Expert Systems, and they are capable of assessing much more complex damage.
Adding home or business assessment through photography, or drone reconnaissance will come, too. Each improvement will speed up claim settlement and make customers happier dealing with insurance companies, instead of being fearful.


Seventy-five percent of customers surveyed want AI insurance advisors. That may seem surprising, but they trust computers to give unbiased advice, and when they say “No” to something, that ends the discussion.
More importantly, most everyone in this country uses smartphones, tablets, or even desktop computers, all of which are ideally suited to ChatBots. Currently most run in messaging format (query/response, query/response,) but some of the more advanced designs can accept verbal inputs and can speak theirown responses (while maintaining a digital record for auditing purposes). AIs are so fast that it could be indistinguishable from a truly knowledgeable person, but if it sounds a bit robotic (stilted) people seem to like it better and trust it more (cf. uncanny valley).

The Added Advantages

Another reason to embrace ChatBots is that they can handle basic customer inquiries, address claims, follow sales leads, address customer insurance coverage questions, and even sell new products.
With AI handling client claims directly, it also makes fraud surprisingly easy to identify. Algorithms can spot fraud-like activity and red-flag any claim for the attention of an assessor. This is a task that ordinarily falls to fallible humans (who can’t possibly catch everything), to the tune of $40 billion of fraud per year (that we know of).

Marketing: The End of the Cold Call

Most salespeople dread the cold call, except for the occasional optimist. That person starts a countdown with the first call (20—No thanks) (19-No thanks) (18-No thanks), knowing that each “no” brings them one step closer to the 1-in-20 who will say “Yes!” But that is basic psychology and is an issue for another day.
Since cold-calling is so time intensive, and few of us relish it, let’s turn them all into warm-calls with AIs. How useful would it be to offer them out-of-country medical insurance, income replacement insurance, or some other financial vehicle because you knew they needed it?
A customer has a digital footprint online. It’s all publically available information, too. An AI could potentially comb through their presence online and assess their risk. It could understand that they take long trips across-country; that they travel to foreign lands; that they engage in risky sports like skiing, mountain climbing, or lawn darts.
If a customer maintains a healthy lifestyle, yoga, running, not smoking, and so on, they may fall into a much lower risk group, and they can obtain a reduction in premiums to reflect their low level of risk to the company. Conversely, those engaging in higher risk activities could be recategorized to reflect the amount of risk they represent.

AIs will be Better at Rating

Actuarial tables and other tools were invented because we couldn’t know each person individually. The time is coming when we can know our customers more intimately.
We’ll be able to let the fellow who has been driving for 30 years without an accident or ticket pay $250 per annum. In turn, we can make the clown with four accidents, two DUIs, and four charges of “stunt driving” pay their full share.
The difference is, if the person has never been entered into the system, has never gotten caught, that doesn’t mean they don’t brag about it online. AIs can identify those who are “getting away with it” for reassessment, or closer investigation.
Many insurance firms now offer discounts if you have a driving-tracking module installed into the diagnostic port of your car. Using GPS, it can record 60 mph in a school zone, or 130 mph in the industrial part of town at 3 am, or any other dangerous activities. Even disconnecting it will leave a pattern of disconnections that can be assessed.

Flip Side

On the other hand, linked to traffic conditions, weather, and whatever other activities will affect driving, such an AI system could feedback to the driver. It could tell them that their planned route will take them into foul weather, icy roads, burst water main, or a 10-car pile-up and direct them away from the danger. Keeping occupants safer reduces the likelihood that they’ll become a statistic, or a new claimant, too.

Will Car Insurance go away?

This is an excellent question! Who pays insurance for self-driving cars? They are coming; have no doubt on that point! Not the passenger (at least not directly); they’d be innocent victims. There is no driver, so maybe the company operating the vehicle?
Here’s the problem: What if they never have accidents? What if the system controlling the vehicles is so sophisticated and flawless that nothing tragic ever goes wrong? Certainly, they’ve shown that they make fewer mistakes than humans—and that is the goal—to have accident-free driving.
Maybe the only people with auto insurance will be those that insist on having a self-drive option, and it would cost $10,000 per year. It might be even more because who knows much risk they would pose among thousands of completely predictable vehicles that communicate amongst themselves and travel in perfect synchronization?
People driving their vehicles could become social pariahs, hated by everyone, for putting so many people at risk by “driving”… The world might be different in the next decade or two with the implementation of AI into everyday life. Letting people be responsible for navigating tons of metal and plastic by hand might be the ultimate social sin!

The Takeaway

Ultimately, how well AI works for us in the insurance industry depends on how we use it. And make no mistake—we will use it. You can create an AI to analyze how people feel about your company (if you’re brave enough) because it gives you the opportunity to make changes to alter that opinion.
You can create AI Agents to mine the internet for data, monitoring hundreds of websites. It can collect, collate, and then present it to you in graphical format, with that one super-ability we all love: finding obscure relationships that would never have occurred to a mere human with limited resources, time, and mental capacity.
Even your call centers will use AI. The ChatBots will take much of the load off of personnel who are currently answering the same basic inquiries again and again. This will free up real people to handle more complex call activities.
Call centers can benefit even more. All hires complete a personality test, and then their activity is monitored for a period to see how they interact with certain callers.
Callers, automatically identified by phone number, would have their profile, built by the AI through data-mining. No the can be connected by the AI to a staff member of a similar disposition, creating an instant rapport. That significantly improves the odds of a customer going away with a problem resolved to their satisfaction and a pleasant memory about talking to the nice tech person that “knew exactly what I needed.”
If you want to learn more, we provide webinars to show you where the future is going and how you can get there. Don’t shortchange yourself. Join us today!