Actionable Insights From APT's Retail Practice

AI Comes to Life in the Insurance Industry

May 5th, 2017 | Posted by APT in Insurance

The robots are coming! According to a recent report from Accenture, 75 percent of insurance executives surveyed believe artificial intelligence (AI) will transform or bring significant change to the industry over the next three years. While insurers are already using some elements of AI, new teams and departments within these organizations are also starting to leverage the technology. As insurers strive to remain on the cutting edge and continue to evolve their use of AI, it is critical that they fully understand how it impacts existing processes before investing in widespread implementation.

Currently, insurers use AI to automate administrative components of different processes, streamlining operations and increasing efficiency by supplementing the efforts of agents, adjusters, and underwriters. Many insurers leverage the technology primarily for tasks like accelerating the claims process by comparing photos of damage to anonymized images of similar damages, to more quickly provide an estimate. Some, like Allstate and GEICO, use robo-advisors or “virtual assistants” to answer basic policyholder questions, while others offer interactive touch displays and virtual real platforms.

As insurers contemplate the future of AI in their organizations, they face a fundamental question of whether it is strategic to leverage these still-developing technologies, which may not yet be fully refined. While there may be an early adopter advantage for swiftly implementing cutting-edge AI initiatives, there could be unforeseen challenges with rollout – in addition to the high upfront cost. Insurers must carefully assess the areas of their organization where it will be most valuable to deploy the technology.

The fastest and most reliable way for insurers to determine an AI initiative’s true overall impact will be to design a test vs. control experiment, where the program is trialed in a subset of the network. Before conducting a test, insurers need to ensure their “test groups” are as representative as possible of the broad network in order to more accurately predict the effect of the program.

Consider an insurer that wanted to introduce technology to provide repair estimates by automatically comparing photos of damage to images of previous, similar damage. Claims adjusters would leverage the program, which would use AI technology, to replace the process of manually scanning photos. Executives may suspect that the program would save time and lead to increased efficiency, but would likely need to verify the impact to justify the cost.

Before implementing the program across the network, the insurer could trial it with some claims and not others. This would allow them to compare resulting metrics like cycle time, customer experience and satisfaction, and associated costs – like the costs of rental cars insurers must cover when policyholders’ cars are being repaired. One consideration may be that the technology is not as accurate in providing estimates as a human would be; however, testing would enable the insurer to whether it would be worth the cost of allocating time to quality assurance for automated estimates.

Overall, leveraging test vs. control analysis would allow the insurer to both unlock insights on the program’s effect on internal processes and achieve a more holistic view of its performance to better understand whether broader implementation would pay off.

While many applications of AI in insurance today are related to administrative tasks in different processes across functional areas, the future of AI will likely extend to more complex tasks and more extensive customer interactions, particularly related to sales and services. The multitude of possibilities for the continued growth of AI makes it all the more important that insurers start critically evaluating and refining their strategies to leverage this technology now, before it evolves from a trendy innovation to a business standard.

Be Sociable, Share!

You can follow any responses to this entry through the RSS 2.0 Both comments and pings are currently closed.