Insurance as an institution is centuries old, the joint-risk concept going back past the medieval era, and today is experiencing disruption and change like none we’ve seen across global sectors in the last hundred years. The gradual introduction of Artificial Intelligence (AI) into underwriting, assessing risk, and detecting fraud at the consumer and insurance agency level will streamline and revolutionize the process, and mitigate the large chunks of manual processing and human error that can save insurers and policyholders time and money at every turn.
Such a foundational change to an industry won’t happen overnight, but as many companies across the world begin to integrate machine learning into their application process, they have noticed a beneficial relationship. “Our ability to actually look at these textual data sources and pull out highly relevant information is greatly increased [with machine learning],” said Andy Breen, senior vice president of digital at Argo Group, in a recent article in Business News Daily. The company is on the cutting edge of using natural language understanding (NLU) to bolster their understanding of the consumer, their needs, and their insurance history.
AI can produce documents and sources in tenths of a second, where a human can further assess the data. This same teamwork is evident when machine learning is used in fraud detection – and the results are staggering. Shift Technology, an AI startup out of France, has processed over 77 million claims with a 75% accuracy rate for detecting fraudulent claims to date. While this number is comparable to human-led data science in the same areas (and might still be a little below the target range for many companies), the accuracy and speed will ramp almost exponentially over the next several years because, quite simply, the robots don’t get tired.
Here is the beautiful symbiosis that will change the DNA of the insurance industry: machine learning can do menial tasks faster and better than any human ever could, and it allows companies to focus their time and efforts back on the product. AI can filter through massive amounts of claims with enough accuracy, while data scientists can update, reiterate, and fine tune the machine in real time. “We’re a ways away from a computer underwriter. We’re really just augmenting humans at this point,” Breen said. In this instance, the claims adjuster, the underwriter, the customer service agent, they all retain their importance and have a great new tool to get all the boring stuff done quicker. Insurance has always been and always will be a human industry. Machine learning can just make us much better at it.
LMA Newsletter of 9-25-23
