Colloquium: Pedram Hassanzadeh

Artificial intelligence (AI) is transforming weather and climate modeling. For example, AI-based weather models now outperform traditional physics-based models for short- and medium-range forecasting, while requiring only a fraction of the computing time. AI “emulators” for long-term simulations of the atmosphere, ocean, and coupled climate system are also emerging, offering new possibilities for reducing uncertainties. At the same time, growing evidence highlights AI models’ difficulties in capturing the rarest (yet most impactful) weather extremes (the so-called gray swans) and learning multi-scale (in time and space) dynamics. In this talk, I will discuss these advances and challenges, as well as ideas for making further progress.