Auto insurers are in the middle of a perfect storm. Performance is suffering as more sophisticated vehicles are proving costlier to repair. Rate increases, designed to help fund innovation and bring combined ratios closer to 100, will be tough for consumers to absorb when we consider inflation-adjusted wage growth has barely hit double digits in the last 40 years.
Another external pressure for insurers is the changing consumer mindset. Consumers are now accustomed to driving experiences themselves, directing all their interactions through their phone. This means every company must meet their demands for service delivery, which generally means self-service, until it doesn’t. Then you need to be ready to deliver personalized service.
Lastly, the promise of advanced technologies, including mobility, AI, chatbots, and telematics, are finally ready for prime time. This readiness cuts two ways: the required technology is finally able to help insurers perform better and deliver better consumer experiences; but, applying these technologies is new, generally not well understood, and can be disruptive.
So, how can these technologies address today’s business challenges and fundamentally change the process in insurance? Let’s look at that.
Technology Change Agents
AI, the Internet of Things (IoT), and mobility will transform insurance. These three capabilities are recreating the world as we see it. And we know this because this is what we’re buying in our personal lives. Think about the first things you see when you walk into a Best Buy. It’s phones, Google home, and the devices that let you turn on the lights remotely or talk to Amazon Alexa. It’s all IoT, delivered through mobility, run by AI.
So, what exactly is AI? Generally, AI is computers that learn and act and make decision like humans. At CCC we have a different view: AI is most powerful when you have machine learning and deep data and logic working together. In our case, we use the collective intelligence of subject matter experts – more than 250K insurance appraisers, estimators, etc. who have connected to CCC – to train the machine to solve very specific problems exceptionally well.
How Does AI Transform Business?
There’s a famous Google cat experiment. Google used visual analytics to train a computer to recognize a cat. It showed its computers millions of pictures of cats, told the computers that’s a cat, that’s a cat, that’s a cat, that’s a cat. Then it showed its computers a whole bunch of pictures that weren’t cats. That’s not a cat, that’s a chair; that’s not a cat, that’s a projector; that’s not a cat, that’s a phone. Eventually the computer learned with a very high degree of accuracy what a cat was.
Now, this is not how most of us learned what a cat was, but don’t underestimate the power of that experiment. AI is changing industry after industry. Here are some examples.
- Transportation. Self-driving cars, self-flying planes, predicting engine failure.
- Manufacturing. Computers or robots are controlling other robots on the manufacturing line, powerful enough to remove components off the line that fail quality tests.
- Network intrusion. Security through fingerprinting, facial recognition, etc.
- Finance. Portfolio management and optimization.
- Healthcare. Computers and AI now routinely screen MRIs and other scans for cancer. Cancer? Not cancer. (Similar to cat, not a cat.) And, they are as accurate as humans.
And even beer. Recently, I overheard a gentleman sitting next to me comment on how great the beer he was drinking tasted. The restaurant manager heard this and offered to explain why that was. As it turns out, some beer kegs today are instrumented with sensors and there’s all kinds of advanced analytics running, including when the beer was first tapped, the kinds of temperature fluctuations the beer went through, and a number of other external environmental factors. The keg from which this glass of beer came, was fresh because the manufacturer recently called the establishment we were in and told them to pour out the old keg; they were sending them a replacement because the one they had was not going to be a good experience for the consumer.
Here’s the point: the beer maker doesn’t sell beer to consumers; they sell to restaurants and middle men. The interesting part of this story is they optimize the consumer experience with their AI and then the value worked all
the way back up the chain. In this and the other examples above, AI is creating trillions of dollars of value and it is enabling all kinds of new experiences for people.
AI in Insurance
It’s now our industry’s turn to put AI to work. What we’re seeing in other industries is now happening in claims. At CCC we’re injecting AI into every key point in the claim process. Here are a few examples.
- First, chat bots. The ability to have AI-driven conversations with customers at scale in real time to guide them through the process when they have an accident.
- Second. Total loss detection. We ask consumers for one photograph and from one photograph our insurance customers, with high accuracy, can determine whether their insureds’ vehicle is a total loss or not. How does it happen? Cat? Not a cat. Cancer? Not cancer. Total loss? Not total loss.
- Next. Damage detection. We get a series of photographs of an accident. The technology detects where the damage is on the vehicle and shows the location and severity of that damage using AI and heat maps.
- Virtual inspection. Instead of sending a team of inspectors out to physically look at vehicles, where they can only do a certain number per day, there is more of a movement to have them use photographs sent over the wire to review in a desktop environment. We are injecting AI into that virtual estimating system. The result? Vastly higher productivity. Your typical estimator can write a lot more estimates in one day using this AI guided estimating tool.
- Auditing. Once the estimate is written, information gleaned from the photographs helps our customers determine if the estimate can be made more accurate and improved.
- Estimating. Those same photos can be used to automatically generate a baseline estimate that can then be edited interactively.
- Lastly, it’s not just about auto-physical damage; it’s also about bodily injury. The use of AI to detect from an accident, the principal direction of force, the delta V, and to predict the likely injuries and their probabilities of all the occupants.
AI Alone Won’t Solve for Everything; Integrating AI into the Claims Process
I’ve seen a lot of promising AI go nowhere. If you can’t take AI and inject it into the process, into the customer experience, into the estimator experience, you have not accomplished anything. We need to integrate AI into the process and the experience. Remember the beer company; it’s about the user experience and if you get that right it drives the value all the way back up the chain.
Powering the Consumer Journey
There has been an accident. The accident has automatically been detected with AI and telematics data. By understanding the rate of change of the vehicle using sophisticated calculations, we know an accident has occurred. No photographs needed; AI and telematics come together to initiate a new claims process, unseating what has been place for a hundred years. Customers won’t call you when they get into an accident; you will know immediately that they have been in and accident and you will contact them.
Next, we’ll go through a chat bot, ask the customer for a photograph, and the AI will be used to help the insurer determine if the vehicle is a total loss or not. In this case the vehicle is repairable. We’ll use the same chatbot to get some additional facts about the accident, ask for some additional photographs, using technology to drive precision because you can’t just ask customer for photographs. You have to make sure you get the right photographs. You need the technology that makes sure that those photographs are useful for the targeted application of finding damage.
Then we ask the driver if he wants to file a claim. If he does want to file a claim and repair the vehicle we can interact with him and show him various shops that are in the geographic area and have the skills to repair that vehicle. He can see the ratings and rankings of these shops provided by previous customers who have repaired their vehicles there. Furthermore, we can offer available times to schedule an appointment to get the car in the shop. These experiences are very similar to those experienced in other domains, i.e. booking a restaurant using Open Table. It’s important, and we’ve taken great care, to build out the estimator experience in a similar way.
So, if you’re not already doing some of this, it’s time to ask yourself where you are in the process. These technologies aren’t going to slow down, but it doesn’t mean you should run with scissors either. There are risks; this is a complicated industry. You’re dealing with business rules of how you work with your suppliers, what kinds of parts you use, labor hours, etc. You are working with regulators. You are working in a legal environment and you have decades of data (representing experience) that you want to maximize for these purposes.
For all its potential, there are risks with applying AI. Here are some key considerations as you embark on your AI initiatives.
- Too much faith in AI. Just because you have good AI does not mean you have the solution to the problem. You need to fuse the AI with the estimating logic, with the regulatory process and with the consumer and estimator experience to really have a solution that’s going to work. It is not just blind trust in AI.
- Transparency. How do people know that what the AI is doing is right or rational? So, the way that they do is by interacting with it and getting more comfortable. That’s kind of what those damage detection photographs and heat maps are all about. They are about the system showing all the participants that the correct damage has been identified. It is transparent. Each step of the way the way the AI gets applied must be thus.
- Trust. This is essential. How many times have any of us seen multiple sheets on the same accident written up the exact same way? Never. The sheets are always different. This is a problem for AI because if we just build up an estimate using AI and show it to anybody in this industry and they say it’s wrong, there is no trust. On the other hand, if professionals interact with those heat maps and usually the heat maps are right, it builds trust. If they interact with a total loss detector and it’s almost always correct, it builds up trust. You get the idea. As this trust builds, they are less likely to say the AI is wrong and more likely to ask themselves what the AI is seeing that they missed. When this happens, human data is training the data, and now AI is also trusted to guide human decision making. This is how an industry gets transformed.
We are at the tipping point; the technologies and process transformations described here are happening. We have many of them being run by customers now. As we embark on this massive-scale change that insurance has never seen, we need to be thoughtful and purposeful. This is a once-in-a-lifetime opportunity; let’s get it right.
Jason Verlen, SVP Product Management, CCC Information Services Inc.
©2018 CCC Information Services Inc. All rights reserved. CCC and the CCC logo are registered service marks; and powering Forward is a trademark of CCC Information Services Inc.
Register or login for access to this item and much more
All Digital Insurance content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access