Russian researchers and international partners have created artificial intelligence capable of forecasting the path of chaotic economic processes. This development was disclosed by the press service of the RNF.
Contemporary science now machines remarkably well. It can model a wide range of phenomena—from planetary orbits to the failure of a brick under stress. While some processes involve randomness, such as the airflow around an aircraft, these elements can be accounted for and folded into a solid mathematical description of flight. Yet social sciences, including economics, contain far more chaotic elements. Markets respond not only to production factors but also to epidemics, shifting cultural tastes (for instance, a demand spike driven by a pop star), and the formation of financial bubbles. Some major events, like the 2008 crisis, emerged from unfamiliar, innovative mechanisms, such as mortgage securitization. Consequently, current economic models fall short of the precision seen in physics or engineering, and researchers are seeking ways to quantify chaotic effects.
Petersburg State University, along with Russian and foreign colleagues, devised a method to identify unpredictable fluctuations in the economy. The approach uses a model that simulates how a society of two generations—young and old—evolves within a given economy over their lifetimes. A pricing component was also explored, incorporating the regional layout of markets linked by a network. This framework has a history of describing the intricate chaotic dynamics of price formation in food markets, including fisheries markets where volatility is a common feature.
Calculations employed evolutionary algorithms and reinforcement learning—areas of artificial intelligence focused on self-improvement through problem-solving experiences. The algorithm not only produces options but also weighs past outcomes to guide future decisions. The project ran on a computer from the Czech National Supercomputer Center in Ostrava, which delivered results in 48 hours; a conventional computer would require roughly 17 years to reach the same conclusion.
At the same time, the authors caution that the method does not offer a complete or flawless answer. Business leaders and policymakers still rely on intuition formed through years of experience. The data generated by the new algorithm should be viewed as one tool among many for human analysts: AI can generalize large datasets and reveal hidden patterns, but human judgment remains essential for interpretation and action.
Ancient biologists observed that moths’ tails act as decoys for bats, a reminder that nature often evolves strategies that blunt predators and shape ecosystems.