A Helping Hand from AI

Harvey, Irma, Katia, and Maria. Houston, Puerto Rico, Florida, and South Carolina. Over the past few months, some of the strongest estimated hurricanes ever recorded (only once before, in 2007, did multiple Category 5 storms make landfall in North America) brought destruction, devastation, migration, and a humanitarian crisis that, months later, is still ongoing.

The estimated damages from Hurricane Harvey alone range from $65 billion to $190 billion, which includes insurance payouts for destroyed property, damage to infrastructure, and economic losses. The thousands of residents affected have faced the daunting challenge of rebuilding, and the last thing on a victim’s mind is filing an insurance claim.

Insurance claims are handled on a first-come, first-serve basis because of limited human capital. After a natural disaster, businesses and residents start the arduous claims process: note-taking, documentation of damages through photographs and logs, receipt collection, and, oftentimes, chasing the claims they are owed in order to avoid missing out on potential payments. Time-honored methods for filing claims have their place, but, in instances of crisis, claims professionals may be able to take advantage of more immediate ways to deliver aid to victims, achieved through emerging automation technologies.

The insurance landscape is being reshaped by automation, or the application of technology to monitor and control the production and delivery of products and services. These developments can include software and hardware tools, and artificial intelligence (AI), which is when computers mimic cognitive functions to find a solution to problems that typically require human understanding.

A report by Tata Consultancy Services estimates that insurance companies will each spend an average of $90 million on AI in the next three years, and for good reason: Adopting these new tools can make processes more accurate, streamlined, and, most importantly, safer for all parties.

Automation technologies like chatbots, visual recognition software, and drones can accurately assess images for houses, siding, rust, and car models without human intervention—and without the potential risks to humans. This value is ultimately passed on to those in need, as insurance companies can pay claims more quickly than ever before and assist a higher volume of customers at any given time. The largest factor in measuring customer satisfaction is the efficiency around the claims process—specifically claim resolution time, according to JD Power. This presents a huge opportunity for insurance companies to not only gain operational efficiency, but also to improve customer satisfaction.

Chatbots

AI with natural language processing solves a lot of problems for insurers, beginning with the customer experience. Chatbots, powered by natural language processing, are a fast and efficient way for victims to reach out to insurance companies so the process of rebuilding can begin. Insureds can text from the convenience of their cellphones and not be beholden to crashed phone lines or after-hours inquiries about policies and billing. Marketing and underwriting also have the ability to see full profiles of customers and use the data to make better marketing decisions and underwriting models.

Visual Image Recognition

Visual image recognition, powered by machine learning, is helping to place customers back into normal and safe conditions. Machine learning is an application of AI, where systems have the ability to automatically learn and improve from experience without being explicitly programmed. Recognition software with learning capabilities can identify items in a photograph, understand the image, and accurately tag and file damage at scale. Through machine learning, AI can identify what is in an image and pre-populate from the information, thus reducing the amount of time that humans need to analyze it, creating huge operational efficiencies.

On-Device AI

Machine learning requires massive computational power—the kind usually found in rooms full of computer servers and only available to those who can connect to the cloud—but half the world still doesn’t have internet access (never mind access during a disaster). This is where AI mobile SDKs come in—where the AI works its magic directly on a phone or other device (also known as “locally”) without needing internet connectivity. This is significant as it allows mobile users to carry out AI calculations on an iPhone or Android device without a connection, making it convenient for users in parts of the world where internet speeds or connections aren’t reliable, or in times of natural disaster when cellular data is down. Mobile SDKs empower claims professionals and inspectors in the field with a mobile device that can capture damage and process information around it in real time without relying on a network connection.
Mobile SDKs not only allow people to run powerful AI applications during a disaster, but they can also improve the workflow that goes into the repair process when partnered with other tools, like drones.

Drones

Drones are now being used to inspect natural disaster damage, speeding up the process and mitigating risk to humans. A typical roof inspection takes a claims professional approximately an hour. Drone inspections take 10-20 minutes depending on the size of the building, and drones can capture video and pictures for processing, providing an estimate for repairs within days, if not hours. With the damage this season’s hurricanes have left in their wake, some areas are impossible to assess by hand. Drones are safer, faster, cheaper, and fit into a suitcase.
There was little precedent for the sheer volume of damage done over the Atlantic Ocean in 2017, and these environmental catastrophes are not over. With the weather cooling and analysts predicting La Niña’s development and influence on weather patterns, we need to remain prepared.

Accidents and disasters happen, and the claims process is complex and regulated, but it’s up to insurance companies, suppliers, and vendors to help victims as best they can. With the right tools, we can all start rebuilding.