Sediments, ice, corals, and trees are just some of the natural storehouses of information that help tell the complicated history of Earth?s climate. Paleoclimate researchers use these ?proxies,? in combination with numerical models, to gain understanding of the magnitudes, rates, and drivers of past climate variability with the goal of informing understanding of current and future change in Earth?s climate system. The Paleo Perspectives on Climate Change (P2C2) program of the National Science Foundation (NSF) has advanced paleoclimate research through proxy development, data-model comparisons, and synthesis work, and has facilitated interdisciplinary collaboration that has contributed to the growth of the field. This new publication highlights discussions at a June 2021 workshop that focused on identifying potential future paleoclimate research directions to further advance understanding of past climate and better inform the public and decision makers about the expected future. Table of Contents Front Matter Overview Introduction Understanding Past Climate Forcings and Sensitivity Glacial, Ocean, and Land Processes and Feedbacks Resolving Regional Climate Change: Advancing and Synthesizing Knowledge How the Paleoclimate Community Can Better Engage on BAJEDI Issues Closing Thoughts References Appendix A: Statement of Task Appendix B: Biographical Sketches of Planning Committee Members Appendix C: Workshop Agenda

The Earth system - the atmospheric, hydrologic, geologic, and biologic cycles that circulate energy, water, nutrients, and other trace substances - is a large, complex, multiscale system in space and time that involves human and natural system interactions. Machine learning (ML) and artificial intelligence (AI) offer opportunities to understand and predict this system. Researchers are actively exploring ways to use ML/AI approaches to advance scientific discovery, speed computation, and link scientific communities. To address the challenges and opportunities around using ML/AI to advance Earth system science, the National Academies convened a workshop in February 2022 that brought together Earth system experts, ML/AI researchers, social and behavioral scientists, ethicists, and decision makers to discuss approaches to improving understanding, analysis, modeling, and prediction. Participants also explored educational pathways, responsible and ethical use of these technologies, and opportunities to foster partnerships and knowledge exchange. This publication summarizes the workshop discussions and themes that emerged throughout the meeting. Table of Contents Front Matter Overview Introduction Emerging Approaches for Using, Interpreting, and Integrating ML/AI for Earth System Science Challenges and Risks of Using ML/AI for Earth System Science Identifying Future Opportunities to Accelerate Progress Closing Thoughts References Appendix A: Statement of Task Appendix B: Planning Committee Biographies Appendix C: Workshop Agenda