Donald R. F. Harleman Lectureship on Artificial Intelligence

The Earth is a complex, multivariate and dynamic network system.  AI offers many opportunities but also challenges to modeling and understanding it. We aim to discover variables relations, derive physically interpretable models, that are simple parsimonious, and mathematically tractable. Machine learning models alone are excellent approximators, but very often do not respect the most elementary laws of physics so consistency and confidence are often compromised. I will review the main challenges and developments in physics-aware machine learning, interpretable AI and causal discovery for Earth system sciences. This is a a collective AI agenda towards developing and applying algorithms capable of discovering knowledge in the Earth system.