Presented on October 29, 2025, by

Melinda Peng
UCCS

ABSTRACT:

Artificial Intelligence (AI), a term coined in 1956, has experienced several cycles of rapid advancement followed by periods of stagnation. A breakthrough occurred in 2016 when Google’s AlphaGo demonstrated that machines could learn independently and surpass human capabilities in complex tasks like the game of Go. Since then, AI and Machine Learning (ML) have developed at an unprecedented pace, revolutionizing nearly every scientific field and industry.

In traditional numerical weather prediction (NWP), it often took the collective effort of the entire community over a decade to improve one-day forecast accuracy. With the advent of powerful GPUs, however, several ML-based global weather prediction models have emerged by training on ~40 years of ECMWF’s global reanalysis data (ERA5). These models have not only matched but, in some cases, surpassed the forecast skill of the best conventional NWP systems while achieving remarkable computational efficiency.

This presentation will provide an overview of AI’s evolution, highlight cutting-edge ML technologies, and explore their applications in meteorology. We will also discuss recent advancements and prospects in this rapidly evolving field.