Analytical Geopolitics and Artificial Intelligence as Hybrid Approach

Anticipating international conflicts is critical for geopolitical risk assessment and early warning. Using the Russia – Ukraine crisis as a case study, this paper presents a hybrid approach that combines theory-driven and data-driven forecasting methods. First, an analytical geopolitics framework identifies structural and agent-based drivers of instability to explain the trajectory toward war. These insights also inform the preprocessing of event data by guiding the differential weighting of conflict events based on their geopolitical salience. Second, we develop an AI -based forecasting model that leverages event data from the Global Database of Events, Language, and Tone and Global Peace Index scores to predict monthly changes in national stability. A Long Short-Term Memory network is trained on these time-series inputs and evaluated against traditional forecasting benchmarks. Results show that the model accurately anticipates key inflection points in the Russia – Ukraine conflict and outperforms classical methods in short-term prediction. By integrating causal geopolitical reasoning with machine learning, this study offers a more robust and interpretable framework for forecasting emerging crises