AI Can Predict Tipping Points for Future Disasters, Scientists Says

A new AI program developed by computer scientists aims to predict catastrophic tipping points in complex systems, potentially forecasting ecological collapses, financial crashes, pandemics, and power outages, Live Science reported yesterday.

Senior study author Gang Yan from Tongji University stated that forecasting critical transitions could help mitigate damage. The research, published in Physical Review X, highlights the challenge of predicting tipping points due to the complexity of systems, often relying on oversimplified models.

The researchers combined two types of neural networks to analyze interactions within systems. They trained the model using theoretical systems before applying it to real-world data, such as the transformation of tropical forests into savannahs. Remarkably, the AI predicted outcomes in a third region using data from two others, even with 81% of nodes unobserved.

Moving forward, the team plans to explore the model’s patterns and apply it to other systems like wildfires and pandemics. They aim to focus on human systems less influenced by behavioral changes to improve prediction accuracy.