Best's Review



At Large
Smarter Tools

Artificial intelligence and machine learning can be buoys in a sea of data.
  • Scott G. Stephenson
  • November 2018
  • print this page

Scott G. Stephenson

Scott G. Stephenson

[Artificial intelligence] doesn’t necessarily imply robots or gadgets, but highly targeted tools that help organize, streamline and automate routine processes.

Channeling the flow of data into the business world has been compared to tapping a gusher of oil—companies are awash in data carrying a potential wealth of insights for those that can handle the volume. The challenge will be to do so without drowning.

Staying afloat may depend on building a digital refinery. Artificial intelligence (AI) is an umbrella term that covers a host of applications, including image analytics, information extraction and speech recognition. AI is helping develop that structure, which may soon lead to greater efficiencies for insurers that adopt the technology.

Contrary to myth, AI doesn't necessarily imply robots or gadgets, but highly targeted tools that help organize, streamline and automate routine processes. Rather than replacing people, advances in AI will likely augment the human element, stripping away repetitive tasks in the same manner as any evolving tech, such as laser welding on an automobile assembly line.

For the record, AI tries to mimic ways that humans reason and approach problems, while machine learning looks to outline patterns in pools of data.

AI and machine learning hold promise for insurance operations, and research is underway to rethink some of the insurance industry's essential tasks. On the underwriting side, one effort is using machine learning to interpret aerial images, extracting data points about roof conditions and siding materials that can help insurers evaluate risks to commercial and residential properties. One result could be better predictive models for property damage tied to extreme weather events.

For claims, we've watched technology transform the adjuster's role. A series of images sent via smartphone from an accident scene can lead to a repair estimate and rapid settlement of an auto claim. But what about fraudulent images? AI studies are working to flag images that have been altered or photoshopped, automatically tagging claims that deserve closer scrutiny.

In other applications, AI studies in voice recognition could help lessen wait times to call centers. With multiple parties on a call and different accents coming into conversations, software is being developed that would identify a caller and anticipate the caller's issue or question based on previous call history. The caller could then be routed to a specialist, eliminating delay and allowing the humans to focus attention on a specific question or solve a problem.

Who's going to be the first to leverage these innovations? The potential disruptive impact from AI and machine learning appears to be monumental within and beyond the insurance industry.

In insurance, we're hearing spirited discussions about AI's role in business operations and portfolio optimization as well as its promise in driver safety.

In a broader sense, these technologies can also act as a lens for insurers to focus on the most pressing problems or data sets, as a microscope for revealing and understanding risk.

Insurers that take a pragmatic approach may find that artificial solutions can deliver real results.


Best’s Review columnist Scott G. Stephenson is chief executive officer of Verisk Analytics. He can be reached at

Back to Home