AI and Rising Conflict Risk_ Predicting the Future

AI tools are enabling more precise and timely predictions of state-based conflicts, moving beyond traditional methods to offer granular forecasts that could inform policy and prevent escalation.

VIEWS Fatality 003 Model

The episode spotlights the Fatality 003 model from VIEWS, launched in 2025, which forecasts state-based conflicts up to three years ahead with monthly updates. For every country, it provides specific probabilities of conflict occurrence alongside expected fatality numbers. This granularity—concrete figures rather than vague trends—allows for targeted risk assessment. Unlike earlier approaches dependent on static historical data, the model processes dynamic inputs for real-time insights, potentially guiding decisions to avert violence.

Satellite Imagery for Urban Unrest

Researchers apply explainable AI to satellite images, achieving 89% accuracy in identifying urban areas affected by riots versus those that are not. This technique pinpoints hotspots where unrest could spread, offering visual evidence of brewing tensions. By making the AI's decision-making transparent, it builds trust for practical use, such as enabling authorities to deploy resources before situations worsen. The method reveals patterns invisible to manual analysis, emphasizing AI's role in rapid spatial monitoring.