Chapter 6:
Challenges With Implementing Artificial Intelligence
Despite AI's indisputable benefits, potential obstacles can prevent carriers and contractors from integrating artificially intelligent solutions into their digital ecosystems.
There are also scenarios that can threaten the long-term success of AI for organizations in this industry.
Below, read about more of some of the most common roadblocks to getting started with AI and also getting the most from the technology once it is in place.
Changing weather patterns requires updated regulations to protect against damage from catastrophes.
Complex Insurance Regulations
Insurers everywhere—as well as their partners—must follow complex and stringent legal requirements in order to remain compliant.
In the United States, for example, each state has its own set of regulations and compliance requirements for insurance carriers. There are regulations for how insurers can build rating systems and pricing models, how they should handle the private data of policyholders, and so on.
These regulations can have a significant impact on insurers, especially when insurance carriers choose to use hazard models to set pricing and establish reinsurance purchasing procedures. In that case, regulators look for disparate impacts before deeming a model that’s suitable for use.
For example, state insurance departments often analyze model results to see if protected classes (those based on age, gender, race, income, etc.) receive higher-than-average scores, regardless of whether their properties are actually at a higher level of risk.
Most states have strict requirements, requiring carriers to submit a rate filing that complies with a multitude of standards to ensure that the models are appropriate for the intended use.
Adding to the complexity of existing processes, states are also reacting to changing weather patterns by constantly updating regulations to better protect people against what appears to be more probable, more significant damage from catastrophes. Taken together, this all creates an overwhelmingly complex, burdensome regulatory atmosphere.
To remain compliant amid the changing environment, carriers and their partners must establish data governance processes that address how to handle all the complex and varying rules in every state in which they have policies in force.
Only after establishing the proper data governance structure—which is naturally a complicated process—can an organization determine whether its environment allows for Artificial Intelligence in their digital workflows.
Even with clearly defined data governance policies, finding a place for AI often involves significant scrutiny; most state regulations still have yet to detail specific AI technology related requirements for InsurTech.
As a result, insurance carriers and other stakeholders in the property insurance ecosphere require guidance both in deciding how to implement technological resources to help them build policies and execute claims tasks.
Since AI technologies process a full spectrum of datasets—some that can include the personal information of policyholders—there isn’t a lot of wiggle room when establishing business strategies that meet all compliance requirements.
Compliance requirements have created an overwhelmingly complex, and burdensome regulatory atmosphere.
Data Silos and Data Quality
The most effective way to operate in the property insurance ecosphere is to approach and execute the entire insurance process as one workflow. For instance, instead of underwriting functions operating independently from other parts, all stakeholders should mold their workflows into one master workflow.
Instead of considering the process as a conglomeration of different business processes, all the various stakeholders must operate as one ecosystem to provide the best, most efficient customer experiences. This approach requires all sides to function as one team with different specialties, using the same, up-to-date, accurate data to power through a roadmap that begins with underwriting a policy, flows to claim initiation, and concludes with claim resolution.
Unfortunately, however, data silos are common among organizations in the property insurance ecosphere. This is especially the case when companies operate without the right data management tools and processes.
Data silos are collections of information controlled by one function or group and untouchable to other stakeholders or business units. These isolated data sets are often stored in one location—digital or otherwise.
For example, underwriting teams often use one tool to store data, while claims teams use another platform that’s incapable of syncing with outside sources.
As a result of disjointed systems for finding and storing data, multiple professionals working on the same claim can easily find themselves working with entirely different, inconsistent information.
Remember the saying “garbage in, garbage out?”
Data silos are a massive culprit of garbage data. This is a significant roadblock to AI implementation, which relies on large, comprehensive sets of consistent and high-quality data.
Data silos are typical, for one, because of legacy technologies that businesses are hesitant to replace. They are also common because insurance professionals do not traditionally function as one collaborative, holistic ecosystem. Until they approach their function, as well as individual data sets, as part of a more extensive operation, stakeholders across the property insurance ecosphere will be limiting the capabilities of implemented AI technology.
To realize the transformational benefits of AI, insurance companies and their business partners must confront this challenge and make significant investments in establishing a data management strategy that enables the secure sharing of accurate, up-to-date information. After all, the “end goal for most AI solutions involves making sense of large amounts of data.”[18]
[18] https://connect.comptia.org/content/research/emerging-business-opportunities-in-ai
Workers need to understand the value of AI and how it will complement—and not displace—them.
Computational Expenses
It’s easy to understand this challenge. Just think of the old adage that “sometimes you have to spend money to make money.” Still, while the concept is easy enough to grasp, the logistics are not always so simple.
Replacing legacy technology with AI platforms, applications, and other systems that support AI can be expensive. Any business that considers such an investment should first carry out a careful evaluation. During this review, it is important to note that, while we can be almost certain that a continued increase in natural disasters (and thus increase in claims) is inevitable in the coming years and decades, we can’t be sure to what degree.
Depending on a company’s specific business objectives, growth goals, and the nature of its customer base (including its geographic location), it may not make sense for a carrier or contractor to invest in specific technologies.
The challenge that then arises is evaluating the consequences of losing a business’ competitive advantage due to a lack of investment in AI technology.
Sometimes you have to spend money to make money.
Company-Wide Buy-in
The success of any technology’s implementation and continued performance depends on widespread usage and understanding of the benefits. Especially with AI, the most significant optimization of business processes will be realized by organizations where everyone knows how to get the best use from the new system.
Getting people willing, let alone excited, is an inevitable challenge in using new systems. This is especially the case among property insurance and restoration professionals, which, as we explained in an earlier chapter, tend to be older.
The property insurance and restoration sectors in general are slow to digitalize, thus increasing the possibility of worker hesitation posing a significant challenge to AI adoption.
With the advent of new AI-based solutions across all industries, there is also a fear that people will lose their jobs to technology that can perform duties and responsibilities more quickly, accurately, and efficiently. The world of property insurance and restoration is no exception. A common barrier to successful AI implementation and usage is getting workers to understand the value of AI and how it will complement—and not displace—them.
Investing the proper time and effort to familiarize your workforce with AI and helping everyone understand how they can complement the technology with their unique human abilities is crucial.
Achieving the Right Balance Between AI and Humans
Too much of anything is never a good thing. At the rate at which AI is evolving, it is not hard for many of us to imagine a world run by robots, void of human empathy and ethics.
As more companies across the property insurance ecosphere embed AI into their digital ecosystems and business operations, organizations will inevitably have to draw a line where the technology stops and where humans have ultimate authority.
There will always be instances when technology cannot process data as a human would. And regardless of how rare these occurrences are, in the insurance world, a single decision at any point in the underwriting or claims processes can have a meaningful impact on the lives of real people (often, your customers).
At the foundation of every insurance-related business, there must be a hybrid of innovative technology-backed workflows and human oversight. Striking the right balance is a critical challenge that every company must achieve in order to ensure that they consistently leverage AI optimally.