Imagine pausing at a stop sign as you drive to work. Suddenly, you are jolted forward as you hear a loud crash. A distracted driver has rear-ended you. You are uninjured, but the back of your car is a mess of crushed metal and broken plastic. Instead of panicking, you pull out your smartphone to access your insurer’s claims app. A virtual claims professional instructs you to talk through the details of what happened. Next, it suggests that you shoot a 360-degree video of the damage to both cars as well as the accident scene. You walk around the cars to capture the footage and quickly upload it. Within a few moments, the virtual claims professional tells you that a claim has been opened and your car is repairable. The virtual claims professional then lets you know a tow truck is on its way, and a ridesharing service is coming to take you to work.
This isn’t science-fiction; it’s the new world of auto claims enabled by artificial intelligence (AI). What the driver in this scenario wasn’t aware of was the host of technological advancements working behind the scenes to create a fast, reassuring, and supportive claims experience. As soon as the driver provided claims details, sophisticated AI algorithms and powerful industry data searches were underway. As the video was uploaded, image analytics processed accident information—including the stop sign and the other vehicle’s skid marks—that supported the assertion that the other driver was likely liable. AI image algorithms extracted the letters, digits, and state information of the other driver’s license plate, and that data kickstarted searches to identify the owner of the vehicle and the corresponding insurance coverage information.
Delivering on the Promise
After years of promises, practical AI applications for insurance are finally here. This new reality has been made possible by an array of technological ecosystems that have created a framework for AI in claims. Consulting firm Strategy Meets Action (SMA) has predicted that 2019 will be the year AI “steps out of the headlines and into business problem resolution” for insurers. AI and machine learning will no longer be toy novelties, but rather workable tools that claims departments can use to boost the performance and results of many processes.
A key driver in the rise of AI has been the advancement of image analysis. The same technology developed for facial recognition in photographs is now being tailored for insurance applications. For auto claims, image-analysis algorithms can recognize car models, identify vehicle parts, detect damage, read license plates, parse documents, and more—all from photos and videos. The performance and reliability of these capabilities have improved to the point where insurers can now feel comfortable launching client-facing deployments.
On the flip side, image-altering software is widely available at low or no cost. This makes insurers understandably wary of taking claimant-supplied images at face value. But here, AI has the answer again. Algorithms for metadata verification and error/noise-level analyses are being honed to detect evidence of image tampering, such as images altered to exaggerate vehicle damage. Photo-clone detection has also improved, including identifying pictures downloaded from the internet and passed off as genuine. These analytical capabilities have matured to the extent that photos that easily fool the human eye can be reliably determined to be fakes. With such controls in place, insurers are increasingly willing to leverage automated image analysis in the claims process.
Next-generation chatbots are also making their way into insurance. They can go beyond initiating a claim; they can ask open-ended questions and use AI to comprehend the answers. Using advances in natural language understanding and sentiment analysis, these chatbots don’t just understand the meaning of words, they also know how a customer is feeling. In fact, chatbots can tailor their responses to show empathy when someone is sad, or appear calm when someone is upset. AI advances in speech analysis and synthesis enabled the development of voicebots that can understand and respond conversationally. That is important because, at the scene of an accident, insureds who are too shaken to type responses may be willing to talk through the events that occurred.
Insurance industry reports of the mass retirement of claims professionals and the struggle to fully replace departing talent have been strong motivators for insurers to implement automation technologies. These capabilities enable insurers to deliver claims services and help maintain their market advantages despite a shrinking pool of experienced claims professionals. But it is important for insurers to determine the right mix of AI technologies and human claims handling to suit their specific business priorities. Claims departments can develop agile processes informed by business rules as well as AI to quickly determine if incoming claims can be fully automated, need a quick review, or warrant a fuller investigation. This automated sorting of claims into no-, low-, or high-touch is the goal of “right-touch” claims resolution.
Automation for Beginners—and Beyond
How do insurers begin? At the simplest level of automation implementation, insurers can leverage chatbots (or voicebots) to support 24/7 claims reporting and perform simple triaging to the right claims professional, whether field or desktop. A more advanced level of automating would be adding system connectivity to enable policyholder collaboration. Designed as phone apps or online portals, these services keep insureds updated on their claims’ progress, contacting them when additional information is needed.
The next level of AI automation would be adding image analysis combined with auto manufacturing information, accident scenarios, and numerous other data sets. With these systems in place, instant claims processing and automated settlements can become a reality. The near-term potential benefits of this approach include reduced policyholder friction, faster cycle times, and optimized claims-handling costs. The long-term benefits are likely to be customer retention from positive claims experiences and a stronger bottom line.
Additionally, an increasing number of vehicles are rolling off assembly lines with sensors that measure a wide range of data, including fluid levels; acceleration; speed; braking; occupancy; weight; and, increasingly, video; light detection and ranging (lidar) systems; and body panel integrity. Thanks to these data-gathering abilities, high-tech vehicles might recognize they have been in an accident even before their drivers do.
Scenarios include customers who are able to set a preference in advance to initiate a claim automatically in case of an accident. In those instances, vehicle telematics can automate claims initiation—for a true instant notice of loss as well as a fully touchless claim. Rich vehicle data will be transmitted via embedded telematics to insurers’ claim analytics systems. Insurers will be able to use sensor and event data recorder information within moments after impact. AI can analyze these critical event details to draw conclusions about the accident—from the likelihood (and type) of injuries to the extent and location of vehicle damage. If serious injuries are predicted, first responders can be alerted. If the vehicle is not drivable, a tow truck can be deployed.
Using comprehensive body and system sensor data sent from the vehicle, analytic systems can build out a comprehensive repair estimate within minutes of the accident. Leveraging vehicle sensor data along with onboard camera data from before the crash, liability assessment analytics can support determinations of causality as well as assignments of negligence where applicable.
Insurers that absorb instant-notice-of-loss reports can plug in their own AI assessment tools to determine whether a claim needs the review of a claims professional. If not, these claims can then be handled as no-touch—saving valuable claims professional hours and sending settlement dollars directly to the insured’s bank account in record time.
This is the year when AI will move beyond the realm of science-fiction to be deployed throughout the life of a claim. The corresponding gains in efficiency, customer service, and improved outcomes are set to revolutionize the claims experience for the benefit of claimants and insurers alike.
SIDEBAR
The Future Reality of Claims Professionals Isn’t Grim
Even with the most advanced technologies currently available, full claims automation will be possible on only 20 to 30 percent of claims in the foreseeable future. The majority will still require the involvement of a claims professional, whether from the desk or in person. With hundreds of thousands of experienced claims professionals retiring, and many insurers finding it difficult to attract talent to the industry, existing claims professionals are encountering an increasing work burden.
By using automation technologies such as AI, chatbots, and robotic process automation, this burden can be significantly alleviated. Claims professionals can be freed up from mundane tasks to focus on more satisfying work that better uses their expertise and judgment. Remote imagery and collaboration tools can save on valuable travel time, allowing claims professionals to handle claims from the convenience of their desks and experience greater work/life balance and workplace satisfaction.