As with other types of online offerings, digital insurance products and services continue to evolve and are becoming increasingly personalized. Customers are being given more control over “self-service” options as companies seek to reduce areas of customer friction. While carriers and intermediaries continue to innovate insurance offerings and platforms in response to market demand, the regulatory framework in many jurisdictions has not yet evolved to contemplate the true manner in which policies are, and will be, sold, marketed or managed.
Policy management and product suitability
The existing regulatory framework does not always permit customers to manage their policies in a customer-friendly manner. Customers expect to be able to access and download their policies (including digital evidence of their coverage on demand), make basic changes to personal information and even make changes to coverage. Despite its seemingly administrative nature, electronic policy delivery may not always be permitted, depending on the jurisdiction or class of insurance.
Renewals and changes to coverage may require a licensed agent or broker, which may be less feasible in a digital setting. Moreover, products must be appropriate, having regard to the customer’s needs and circumstances. Traditionally, a suitability assessment would be achieved through a discussion between the customer and the insurance broker or adviser in person or by telephone. Customers today may wish to have the option to speak with someone over the phone, but they typically do not want that to be mandatory.
Although it is possible that certain basic insurance products may not require an in-depth discussion and could be satisfied via the use of online questionnaires, more technological progress (and regulatory change) is needed before customers may rely entirely on the use of AI for more complex products.
The disconnect between the speed and efficacy customers expect when binding and dealing with insurance products and the expectations that regulators have in respect of customer protections only seems to widen. Virtual advisers, which combine social, digital and customer-provided data to offer a personalized shopping experience, are gaining popularity in the United States. Meaningful adoption in Canada may be slow, given the use of virtual advisers would need to be combined with the expertise of a live agent to satisfy the requirement that customers are able to interact with a licensed broker or agent to pose questions relating to the insurance coverage applied for at any point in the online journey.
As well, the provincial licensing requirements that are applicable to insurers and their agents/brokers remain applicable in the digital space. Agents or brokers licensed to sell insurance products in Alberta but not in Ontario should not respond to coverage questions or sell property insurance to an Ontario resident for a property in Ontario, for example. This may introduce hurdles for insurers and intermediaries who are attempting to launch national distribution programs.
Artificial intelligence
U.S. state regulators are participating in discussions relating to artificial intelligence (AI), open insurance and other tools to continue promoting innovation. Canadian regulators are also beginning to focus on the regulation of AI and whether the insurance regulatory framework has kept up with industry changes arising from new technology.
Advancements in AI have contributed to the personalization of financial products and services. Customers interact and share data with financial institutions and other entities in a number of different ways. Companies may monitor a customer’s user behaviour on its website and consider contextual data to determine the customer’s interest in and need for additional products or services. Customers are also sharing more information and data on social media. While individual data points may seem innocuous or unrelated to a customer’s insurance needs, when analyzed with other data points, they can be predictive of an insured’s risk rating, product needs, suitability and/or financial means.
Large language models (LLMs) are a type of generative AI trained on large amounts of data. LLMs are able to understand and generate language and other types of content and may be utilized for various tasks, including language translation, text summaries, responding to questions, creative writing or code generation.
However, LLMs may make incorrect statements (known as “hallucinations”). There are examples of consumer-facing chatbots indicating that products or services were available at certain pricing or with discounts when, in fact, they were not. In addition, bias may exist in the initial training data, the algorithm or the predictions produced by the algorithm. The initial data may not represent a sufficient number or degree of viewpoints or may contain stereotypes or other prejudicial information.
Although the use of AI tools like LLMs will undoubtedly permit insurers and intermediaries to market and sell insurance products in a more personalized and consumer-centric manner, users of such technology must monitor the relevant inputs and outputs associated with such tools to ensure that information provided to customers is accurate.
In addition, carriers should ensure that key information, such as premium rates, is not generated based on data that is prejudicial or insufficient. This requires specialists who can detect bias in the models and who have the requisite skills to correct the models and promote algorithmic fairness during training and model development. Carriers also need to understand the limitations of the models they select and deploy. There is currently no true regulation of AI in the context of insurance distribution, so carriers and intermediaries are left with no choice but to substitute existing regulatory principles until there is clarity from the regulators as to how and when AI tools may be utilized.
Conclusion
The continued evolution of insurance distribution will eventually drive change and reshape the regulatory framework to recognize the commercial reality of how customers are purchasing insurance. Until we close the gap between what the insurance purchase experience could be versus what it is currently allowed to be today, insurers need to determine how to deliver on the promise of being “not your parent’s” insurance company while colouring within the antiquated lines of the applicable laws and regulations.
This is the final part of a two-part series. Click here to read Part one: Mind the gap: Compliance in the era of digital insurance distribution.
For more information on this topic, please reach out to the author, Marisa Coggin.
This article was originally published on Law360TM Canada (www.law360.ca), part of LexisNexis Canada Inc.
