Because of rapid technological advances in sensor development, computational capacity, and data storage density, the volume, velocity, complexity, and resolution of Earth science data are rapidly increasing. Data mining, machine learning (ML), and other statistical regression approaches, often referred to collectively as artificial intelligence (AI), offer the promise for improved prediction and mechanistic understanding of Earth system processes, and they provide a path for fusing data from multiple sources or platforms into data-driven and hybrid models, employing of both process-based and deep learning elements. Prospective opportunities for using an AI framework to integrate a wealth of in situ measurements and remotely sensed observations will be presented. As an example, ML-based models of plant stomatal conductance and plant hydraulics can be developed to produce a hybrid process-based/ML-based land model within a global Earth system model with the aim of reducing uncertainties in predictions of soil moisture, plant productivity, and carbon assimilation. Carefully designed hybrid models could improve accuracy of predictions and better inform choices of climate mitigation and adaptation strategies. A variety of environmental characterization, uncertainty quantification, and model prediction approaches will be described, and strategies for developing a new generation of ML methods on high performance computing platforms to advance Earth and environmental system science will be presented.