AI-Oriented Risk Assessment Models in Emerging Insurance Markets: The Case of Uzbekistan

Purpose- The purpose of this study is to examine the transformation processes within insurance markets in developing countries, with a particular focus on the growing influence of artificial intelligence (AI) technologies. While AI is reshaping mature insurance markets in developed economies through automation and predictive analytics, its adoption in developing nations remains uneven and understudied. This research specifically aims to assess the feasibility of implementing AI-oriented risk assessment models in Uzbekistan – a country with a rapidly ex-panding but institutionally immature insurance sector. The study also seeks to propose a phased methodology for AI adoption that navigates regulatory gaps, infrastructural weaknesses, and human resource limitations. Methodology- The study employs a case study approach centered on Uzbekistan, combined with econometric modeling. Specifically, a binary logit model is used to assess factors influencing insurance companies' readiness to adopt AI technologies. Key independent variables examined include firm size, years of digital experience, level of regulatory compliance, access to external technical support, and prior investment in data management systems. The authors also draw upon global best practices from countries that have pioneered low-cost, scalable AI solutions under resource constraints. These international experiences are systematically contrasted with Uzbekistan's specific limitations, including fragmented regulatory oversight, limited cloud computing infrastructure, and a shortage of data science talent. Findings- The analysis reveals that readiness for AI adoption is significantly positively associated with prior digital investment and external partner-ships. Conversely, regulatory ambiguity and staff resistance act as substantial barriers. Unlike developed markets where extensive historical data facilitates AI integration, Uzbekistan's insurance ecosystem faces significant gaps in data standardization, technological readiness, and institutional coordination. However, the findings demonstrate that these constraints do not preclude AI adoption altogether. The logit re-gression results empirically validate that a carefully calibrated, context-sensitive approach is required, moving from pilot projects in low-risk product lines (e.g., crop or micro-insurance) to full-scale AI integration. Conclusion- Based upon the analysis and findings, it may be concluded that even institutionally constrained markets can harness intelligent technolo-gies, provided that implementation is phased, context-aware, and supported by targeted policy interventions. For policymakers, the results offer evidence-based guidance for shaping digitalization strategies, including regulatory sandboxes and tax incentives. For insurance com-panies, the proposed phased methodology provides a concrete roadmap. Ultimately, the article contributes to the growing literature on AI in development finance by demonstrating that constraints do not preclude adoption – they necessitate careful calibration.