Marko Sumina on AI in Insurance

My transition to Adacta as a Business Analyst was driven by a profound aspiration to apply the machine learning knowledge acquired during my studies to innovate within the insurance domain. Joining the IT industry seemed the most fitting avenue to achieve this goal.

The insurance industry holds significant potential for the effective application of AI. However, it's crucial to establish the foundational building blocks before delving into model development and testing ideas. Initiating the process requires well-structured and meticulously designed data storages, followed by the implementation of an AI infrastructure. Only after these steps are in place can you effectively apply algorithms to address specific business use cases.

AI finds various applications in the insurance sector, with a primary focus on claims-related processes. This emphasis is due to the prevalence of paper-based and insufficiently digitized claims procedures. Another contributing factor is the direct impact that the efficiency of underwriting and claims processing has on a company's profitability. Beyond claims, AI is also applied in non-claim-related areas such as enhancing policyholder loyalty, upselling and profiling, metadata analysis, IoT applications, chatbots, telematics, and agent matching.

Always consider the specific business problem you are addressing before collecting data and testing your ideas. Prioritize the end user experience to ensure usefulness and presentability. Your work is truly complete only when end users can make actionable decisions based on the solutions you've implemented.

AI adoption in the insurance industry has progressed relatively slowly compared to sectors like banking. Despite significant investments in insurtech startups, the outcomes have been limited. This is primarily due to challenges such as the insufficient domain knowledge of startups focusing on generic AI applications. Additionally, the adoption hurdles are exacerbated by the presence of large legacy systems within insurance companies that require redesigning to align with these specific purposes.

Embracing these tools is highly beneficial as they expedite the research process and alleviate concerns associated with widespread AI adoption. Comparable tools are already in use for client-related and internal company chatbots. They aid clients and employees in searching, engaging in conversations, and applying Large Language Models (LLMs) to enhance their work processes.

We've primarily delved into the potential applications of these technologies for internal purposes, particularly in the realm of intelligent document search and coding assistants for our developers. However, we continue to explore new possibilities on a daily basis.

For a business analyst, it serves as an excellent aid by offering a tool to comprehend complex subjects more rapidly than was previously achievable. Being an international company, there are substantial advantages in using it to overcome potential language barriers as well.

Predicting anything that AI cannot replace is challenging. The current level of adoption and research is unprecedented and was not anticipated nearly a decade ago, surprising even the most optimistic individuals in the field. Even aspects inherently considered human, like creativity, are gradually being replaced by AI-related applications.