"Long-Term Autonomy: Learning What Matters When" presented by Kagan Tumer, Professor of Robotics, College of Engineering. AI systems face unique challenges when deployed in open-ended, long-term real-world tasks. Unlike games like chess, real world problems do not have a well-defined concept of "win," nor do they always have a clear "end" of the task. In this talk, we'll discuss how long term autonomy requires a new paradigm, one focused on determining "what matters when" rather than how best to achieve a narrow task.