Gallagher Survey: AI Adoption High, Data Risks Remain

- A related Gallagher survey of U.S. business owners found that while 81% plan to maintain or increase AI investments, their top concerns include AI-generated errors or "hallucinations" (34%), data protection violations (33%), and legal liabilities from AI misuse (31%). - From an actuarial perspective, the core challenge of AI governance is managing model risk, reputational risk, and regulatory risk. Actuaries are seen as uniquely positioned to lead AI governance by applying principles of bias assessment, explainability, and model validation to AI systems used for pricing and underwriting. - To manage data risks at an enterprise scale, companies are adopting platforms like Snowflake, which provide a unified architecture for data and AI. This approach brings AI capabilities directly to the governed data, aiming to prevent the security gaps and data duplication that can occur with fragmented systems using external LLM APIs and third-party vector databases. - In consumer retail, fashion brands are using AI for more than just personalization; companies like Zara and Shein apply machine learning to optimize inventory levels and forecast demand, which helps to reduce overproduction and waste. - Big tech companies are pursuing divergent hardware strategies for AI, with Apple reportedly developing on-device AI through smart glasses and camera-equipped AirPods, a contrast to the massive data center infrastructure investments being made by Google, Meta, and Amazon. - Beyond data protection, the most common barriers to enterprise AI adoption are a lack of employee AI skills (35%), the difficulty of integrating AI with legacy systems (29%), and poor data quality (29%). - Within the insurance industry, AI-centered "InsurTechs" are attracting significant investment, accounting for 63.4% of all funding deals in the third quarter of 2024. The primary focus of these investments is on central business operations, particularly modernizing policy administration systems.