Purposeful AI

AI To The Rescue

In the highly polarized debate on whether Artificial Intelligence (AI) is the ultimate evil or the greatest good, the latter argument got a shot in the arm during the recent Hurricane Harvey. A teenage girl ill and trapped in her home managed to contact the Coast Guard rescue team through none other than her personal digital assistant.

In the aftermath of Hurricanes Harvey and Irma, the media is full of poignant stories about how another application of AI, the drone, is proving indispensable in rescue and relief efforts. Less known however, is the role that drones and other types of AI can play in another operation associated with a natural disaster – the assessment and disbursement of insurance claims.

Harnessing drones in a natural disaster has enormous implications given that their scale has been consistently rising. In the 1980s, a billion-dollar natural disaster was experienced 2.7 times a year, on an average. In the 1990s, that number went up to 4.6, and in the 2000s, climbed to 5.4. Estimates are that Harvey and Irma have destroyed property and automobiles valued in excess of US$290 billion, and done incalculable damage to life and livelihood.

For the victims of natural disasters, insurance money is very often a vital lifeline. Unfortunately, with the scale of damage caused by Hurricane Harvey and the fact that it was closely followed by Hurricane Irma, means human claims adjusters could take years just to assess the scale of damage to property. Here, any mechanism that expedites the process and allows insurance companies to pay out claims at the earliest would be immensely valuable.

The drone is proving to be that mechanism. It takes high quality images of affected locations that claims adjusters can view in real-time on their smartphones, or download onto their company database for future reference. Machine learning models study these visuals to locate and identify the extent of damage to property. This information is factored along with data, such as amount the property is insured for, age of the property, damages covered by the policy, etc., to arrive at the admissible claim. The process can be completed within a few hours, with minimal human intervention, and with a high degree of accuracy.

Besides improving the speed and efficiency of claims settlements, drones can also mitigate insurance fraud and malpractice. When Superstorm Sandy hit the US in 2012, it was alleged that the investigator for a particular insurance company changed the conclusion of an engineering report from flood damage to “preexisting damage” without the knowledge or approval of the original engineer. This resulted in hundreds of denied claims being reopened. Had there been drones used for assessing the damage, the arguments could have been supported by more photographic evidence.

A combination of drones, sensors and AI applications can also minimize the risk of injury to claims adjusters by taking over inspection of hazardous spaces.

Going forward, we can expect AI - through smart homes - to be involved much earlier in the insurance value chain, especially in a disaster scenario. For instance, machine learning applications would analyze weather patterns to predict the severity and likelihood of hurricanes well in advance, to not only enable smooth and early evacuation but also allow the affected population to secure their property. Insurers would increasingly use AI solutions to improve underwriting and pricing practices. And relief agencies would send out rescue robots to bring people to safety.

PwC estimates that drones, by virtue of their contribution to risk monitoring, risk assessment and claims management are already worth some US$6.8 billion to the insurance industry. A number of insurance giants, including Travelers and Allstate, have instituted robust drone programs as part of their risk assessment and claims management processes. With their potential to assess damage faster, shorten the claims settlement process, root out fraud, and more importantly, improve customer experience, drones and other types of AI hold irresistible appeal for the insurance industry. This is the chance for a business that has somewhat lagged in cutting edge technology adoption to alter its course.