The Financial Times recently highlighted that the improvements in artificial intelligence (AI) that have boosted weather forecasting could also be applied to model political and geopolitical crises. The piece asked: if one-week weather forecasts can now be more accurate, why couldn't AI provide more reliable predictions about wars, international actions or other crises?
Background
The reasoning is that the best AI-based one-week weather forecasts have improved significantly; in many cases they now reach accuracy levels comparable to three-day forecasts from around 1980, a time before widespread AI use. Weather is scientifically a chaotic system, yet AI has improved forecast quality through better model synthesis and data analysis.
Political and social events, however, are driven largely by human decisions and feedback loops, which pose different challenges. At the same time, these events contain patterns that AI can detect — patterns that human analysts might miss.
A concrete example: Vico's prediction
A recent illustrative case involves the AI system Vico. On February 19, Vico assigned an 89 percent probability that the United States would take military action against Iran by March 31. The action occurred on February 28. This example shows that AI can sometimes detect signals and correlations earlier than traditional analysis.
Why this matters
Together, the Vico example and the improvements in meteorological forecasting suggest that AI could become a more important tool for modeling crises and conflicts. This does not imply these forecasts are infallible: political events remain complex and subject to feedback, so uncertainty persists. Still, AI increasingly offers ways to supplement conventional analysis and forecasting methods.
Debate and applications
In discussions about these developments, commentators often mention prediction platforms such as Polymarket for comparison: AI models and prediction markets approach uncertainty differently. Prediction markets reflect the collective views of participants, whereas AI models search for patterns across large datasets. A combination of both approaches could provide decision-makers with a richer information set.
Conclusion
The progress of AI-driven forecasting — as illustrated by advances in meteorological models — may enable better modeling of political and security crises. The Vico case demonstrates that such systems can sometimes predict specific events accurately, but the inherent complexity of political decision-making means significant uncertainty remains. Going forward, combining AI models and prediction markets may yield the most informative perspectives for policymakers.
