The AAAI 2023 Spring Symposium on AI Climate Tipping-Point Discovery (ACTD) included researchers from a number of disciplines that came together to better understand this new emerging area of research that is an integration of Artificial Intelligence (AI) and traditional climate modeling methods to better understand climate tipping points. Critical tipping points of concern exist in Earth systems, including massive shifts in ocean currents, cryosphere collapse, forest dieback, and permafrost thaw. A growing concern among climate tipping point research is the cascading effects across tipping points. There are many urgent research questions that were addressed during this event, to better understand these rapidly changing and interconnected global systems. Ideally new research in this area could accelerate scientific discovery by overcoming existing limitations in state-of-the-art climate modeling approaches used for tipping-point discovery. Discussions were intended to go beyond the climate community and touched on how climate tipping-point discovery methods could inform discovery within other systems such as social, political, and economic systems. The keynote speakers together provided a comprehensive background of the challenges of current climate tipping point problems and modeling techniques, highlighting specific systems such as the oceans, ice, and forests. Talks described the current state of climate intervention methods, challenges with feedback mechanisms across systems, and emergent behavior in climate systems. AI methods to overcome current modeling challenges were presented by keynote speakers, including methods for better representing physics in climate models, deep learning methods for tipping point discovery, and new methods for intervention. New deep learning-based phase transition methods for early warning signals were highlighted and compared with deep learning bifurcation methods. Large scale climate modeling techniques using various types of neural operators were presented, which yielded compelling results for non-trivial use cases. Detailed discussions focused on the use of AI to support the study of coupled systems, such as oceans and ice, and to gain a better understanding of emergent properties of these systems. Among paper and lightning talks, researchers outlined multiple applications of AI that could support climate researchers to better understand their respective domain both in terms of more extensive discovery and decreased computational load. Methods described in terms of learning to identify bifurcations and tipping point dynamics included the use of Koopman operators and deep learning networks such as Long Short Term Memory (LSTM) type networks and convolutional neural networks (CNN). A generative adversarial network called TIP-GAN which uses an adversarial game to learn tipping points was described, which works with a neurosymbolic model, providing a way for climate researchers to ask natural language questions of the model. Correction based models to account for bias and extreme events included the use of DeepONet neural operators and LSTM methods. In addition, interesting research was presented related to climate intervention methods, including a marine cloud brightening technique and a stratospheric aerosol injection method. Specific use cases were also described, including using virtual reality to better understand climate systems, using a CNN and satellite data to learn road transportation for better estimation of emissions, and a U-Net deep learning model for arctic sea ice operations. Finally, panel discussions addressed topics related to the future of climate tipping points, intervention methods, how to tackle cascading tipping points, and potential funding avenues to continue support of this line of research. The initial goal of this symposium was achieved—to explore how traditional climate modeling methods and AI could be combined for climate tipping-point discovery. The community of researchers working at this intersection of AI, dynamical systems, and climate science have begun to shape a vision of this important field of ACTD. The papers of the symposium will be published as a AAAI Press Technical Report and as part of a future open access proceedings. Authors: Jennifer Sleeman and Jay Brett with the Johns Hopkins University Applied Physics Laboratory Authors Emails: [email protected], [email protected] Chair: Jennifer Sleeman Organizing Committee: Jennifer Sleeman, Anand Gnanadesikan, Yannis Kevrekidis, Jay Brett, Themistoklis Sapsis, Tapio Schneider Program Committee: Maria Fonoberova, Aniruddha Bora, Alexis-Tzianni Charalampopoulos, Thomas Haine, Ignacio Lopez-Gomez, David Chung, Mimi Szeto, Chace Ashcraft, Anshu Saksena Approved for public release; distribution is unlimited. This material is based upon work supported by the Defense Advanced Research Projects Agency (DARPA) under Agreement No. HR00112290032. The author declares no conflicts of interest. Dr. Jennifer Sleeman is a Senior Artificial Intelligence Researcher at Johns Hopkins University Applied Physics Laboratory. Dr. Sleeman has over 20 years of experience as an Artificial Intelligence researcher and received her Ph.D. in 2017 for her work Dynamic Data Assimilation for Topic Modeling (DDATM). Dr. Sleeman's work includes AI for climate and weather and has used deep learning for climate tipping point discovery, extreme weather pollution events, as well as performed data assimilation of climate science reports from a 30-year period. She holds a patent in Deep Model Translation and Generation, and has been a major research contributor in the area of deep learning for image recognition, knowledge graph representation and natural language processing. Dr. Jay Brett is a Senior Oceanographer at Johns Hopkins University Applied Physics Laboratory with expertise in numerical modeling of physical and biological processes of the ocean. They studied physical oceanography in the Massachusetts Institute of Technology-Woods Hole Oceanographic Institution Joint Program, receiving their PhD in 2018 for their work on numerically modeling three-dimensional ocean eddies and gyres and analyzing the stirring and mixing therein. Dr. Brett will provide ocean physical modeling expertise to the Use Case Modeling, Evaluation, and Data Acquisition groups.