The Machine Learning and the Physical Sciences workshop aims to provide an informal, inclusive and leading-edge venue for research and discussions at the interface of machine learning (ML) and the physical sciences. This interface spans (1) applications of ML in physical sciences (ML for physics), (2) developments in ML motivated by physical insights (physics for ML), and most recently (3) convergence of ML and physical sciences (physics with ML) which inspires questioning what scientific understanding means in the age of complex-AI powered science, and what roles machine and human scientists will play in developing scientific understanding in the future.