The Landscape of Insurance in 2024: A Shift in Focus
The purpose of the insurance industry has always been to make risk manageable and thus reduce financial uncertainty. Therefore, trust is at the core of the relationship between any insurance actor and its customers, as it is impossible to believe that someone or something will support you when things go wrong unless trust has been built upfront. In this respect, we may say, metaphorically speaking, that insurance is selling trust and peace of mind. It is the intrinsic nature of the business to exercise caution and balance the internal and external risks when it comes to changing how business is done.
Still, in a world where speed is the only constant, remaining unchanged can become riskier than change itself. Many factors drive change, including technological advancements, regulatory changes, the shapeshifting of traditional risks, the emergence of new risks, and evolving customer expectations.
Profitability is a natural priority for insurance market players, as it provides strength and stability and projects trust. The tremendous potential of technology to provide more than marginal gains drives companies closer to the aspirations of their customers (including new generations), improves the end-user experience, and makes new risks insurable.
Understanding risk, making it manageable, and creating affordable products are at the core of business innovation in insurance, which has always been a data-centric industry that evaluates trends in data and calculates probabilities. That’s why anything that makes sense of data better, faster, and cheaper opens various opportunities to create better customer experiences, become more profitable and inclusive, and open new markets with new products.
The Power of Generative AI
Believe it or not, the first AI algorithms, which are the basis of what’s going on today, have been around since the 1950s. So, why has it taken us 60-70 years to get to where we are? The answer is simple and complex at the same time: the democratization of computing resources. Any data analysis algorithm, no matter how efficient, needs two things: electronic data (the more, the better) and computing power (depending on the data, but a lot). Trends in data gathering and generation have been creating huge amounts of electronically available data. At the same time, computing power has developed exponentially; not only has it become cheaper, but more importantly, it is fractionally available via the cloud. That’s why powerful analytics or AI is unavoidably linked to cloud resources, moving from an own-it-all approach to a rent-what-you-need one, thus becoming more affordable. In simple terms, to get the power of analytics at a decent cost, you must first embrace the cloud.
What makes Gen AI different? It is the ability to generate rich content based on what it learns. By rich content, I mean that having a conversation with a bot, summarizing a large document, or generating code based on human requirements are all possible because of Gen AI’s intrinsic ability to understand context and semantics. The ability to generate search results based on conversations can also be added to the list.
Here are a few use cases.
Content Generation: From generating responses to customer inquiries in call center analytics to creating personalized user interfaces for websites, content generation enhances customer engagement and streamlines operations. Thus, a quote and bind journey can be transformed from tedious Q&A interrogations into a conversation. For example, Geico’s AI-powered virtual assistant, “Kate,” helps customers with policy information, billing inquiries, and other service requests through its voice and chat interfaces. This enhances customer engagement by providing quick and accurate responses (Insurance Journal).
Progressive uses chatbots and AI to assist customers with claims filing and policy management, reducing wait times and improving the overall customer experience (MMA Insurance).
Summarization: This feature summarizes customer support conversation logs for call center analytics, documents (e.g., financial reporting, analyst articles), and social media trends to provide concise, actionable insights. This includes summarizing the documents filed with a claim to detect missing information or discrepancies, leading to lower costs of claims relating to handling and fraud, while improving customer experience.
Lemonade uses AI to process claims through its virtual assistant named Jim. Jim can process simple claims in seconds, analyzing various details and cross-referencing them with policy information to reduce human intervention and speed up the claims process (CBIZ).
Another case presented in the media is Anthem. Anthem utilizes AI to detect fraudulent claims, rapidly analyzing patterns and anomalies in claims data more accurately than traditional methods (Digital Insurance). The ability to analyze and summarize data is not only useful in claims management but also in underwriting, especially when it comes to risk selection, risk assessment, pricing and personalization.
Swiss Re uses algorithms to analyze vast amounts of data from various sources (e.g., social media or satellite imagery) and get insights regarding risk parameters. Not only does this improve risk assessment and underwriting accuracy, but it also enables the delivery of personalized and precise insurance products (Insurance Journal).
In the case of Allstate, AI is used to enhance the underwriting process, accurately predict risk factors and set premiums that are better aligned with the actual risk (MMA Insurance).
Code Generation: This capability improves data handling and software development by converting natural language into query programming languages, like SQL (or vice versa) for telemetry data, querying proprietary data models, and generating code documentation.
Semantic Search: With this feature, you can search reviews of specific products and services, discover information, and mine knowledge for deeper insights and better decision-making.
These capabilities can be combined to create multiple model use cases, such as end-to-end call center analytics (classification, sentiment, entity extraction, summarization, and email generation), customer 360 (hyper-personalization using timely summarization of customer queries and trends, search, and content generation), or business process automation (search through structured and unstructured documentation, generate code to query data models, content generation).
John Hancock uses predictive analytics to tailor life insurance products. By analyzing health data from wearable devices, they offer policies that incentivize healthier lifestyles (Digital Insurance).
A Phased Approach to AI Implementation
The insurance industry is ripe with opportunities for AI implementation. However, to ensure success, it must be approached strategically, choosing use cases that are likely to succeed. For instance, some applications, like real-time numerical calculations for actuarial models or complex risk model replacements, may not be suitable for generative AI capabilities. Insurers would also do well to avoid areas with significant regulatory implications, such as risk modeling, and automated underwriting, where better applications would be to create augmented co-pilot-like experiences to help people working in these areas (underwriters, actuaries) get insights where they need them and when they need them.
A good approach when looking at leveraging AI is to land and expand gradually by starting small and increasing the scope as the endeavor becomes successful. Analyzing policy documents, summarizing claim reports, processing forms for policy applications, and generating content to communicate with customers are excellent starting points. Furthermore, understanding historical claims data and the basis of previous underwriters' decisions can also provide significant value.
Considerations for the AI Journey
The insurance industry has started to embrace AI in impactful ways, from claims processing to fraud detection, underwriting, and risk management. Industry leaders leveraging AI have witnessed better efficiencies and product innovation.
However, whether AI will continue to be successful depends largely on how insurers scale AI to address their business strategies, regulatory requirements, and customer needs. They will also have to address the growing concerns surrounding data privacy, algorithmic bias, and lack of AI transparency.
Ultimately, the answer to whether AI becomes the great equalizer, democratizing access and pricing, or whether it inadvertently creates new forms of digital divide lies squarely with the insurers themselves. By charting a course that balances innovation with inclusivity, insurers can ensure that AI is a tool for progress.
