Explore the benefits of acknowledging uncertainty in climate system analysis and modeling using machine learning in this conference talk from the Machine Learning for Climate KITP conference. Delve into state-dependent predictability, challenges in regression problems, and controlled abstention networks. Gain insights into how big data and machine learning algorithms can advance climate science, enabling detailed analysis and potential causal inferences. Discover how this interdisciplinary approach can address complex climate questions and inform future predictions at regional and local scales.