Insurers Urged to Clean Legacy Data for AI
A post from consulting firm DOOR3 emphasized that cleaning and unifying legacy, siloed data is the essential first step for insurers to successfully implement AI. The firm argues that focusing on data hygiene, accessibility, and utility is foundational to improving risk selection.
- Recent industry research indicates that 74% of insurance companies still rely on outdated legacy systems for core functions like pricing and underwriting, which significantly hinders the adoption of modern AI capabilities. - To bridge the gap between AI model development and production, many insurers are adopting MLOps (Machine Learning Operations) frameworks; these practices focus on creating reproducible, automated pipelines for data validation, model training, and monitoring to prevent issues like model drift. - Actuarial bodies like the International Actuarial Association (IAA) have released specific guidance on AI, emphasizing the need for robust governance, model testing, and comprehensive documentation to ensure transparency and manage risks associated with AI-driven systems. - Generative AI is being used to accelerate the modernization process itself by analyzing old, undocumented legacy code and automatically generating structured documentation, which helps engineers understand and migrate complex systems. - In contrast to the siloed, batch-processing nature of legacy systems, modern data lakehouse architectures are being adopted to handle the streaming data from sources like telematics devices and claims platforms, enabling real-time fraud detection and dynamic pricing. - The high cost of inaction is a major driver for change, as maintenance of legacy systems can consume, on average, 70% of an insurer's annual IT budget, limiting funds for innovation. - Drawing inspiration from the consumer sector, insurers are looking at how retailers like Nike and Swarovski use AI to deliver hyper-personalized customer experiences, which depend on unified, high-quality data to power recommendation engines and analyze customer behavior. - For data engineers in the NYC area, the demand for these skills is reflected in a growing number of local meetups, such as the "NY AI Engineers" group, and available roles at companies ranging from startups to major carriers like New York Life Insurance Company.
