A new study demonstrates that artificial intelligence models can generate highly accurate weather forecasts faster and more efficiently than conventional physics-based systems. The research, conducted by a team at the University of California, Berkeley, shows that AI models trained on decades of historical weather data can predict key atmospheric variables like temperature and pressure with comparable or superior accuracy to established numerical weather prediction (NWP) models. These AI systems complete forecasts in minutes rather than hours, requiring significantly less computational power. While the technology shows immense potential for rapid, localized predictions and climate scenario modeling, experts caution that further testing is needed for extreme weather events and long-term reliability. The findings suggest a hybrid future where AI augments traditional forecasting methods. Read the full article at https://sciencedaily.com/releases/2024/05/240507150123.htm.