Recent years have witnessed tremendous activities of Artificial
Intelligence (AI) and deep learning (DL) in our lives. These activities have
traditionally belonged to well-trained human beings from object detection, voice
recognition, self-driving cars to decision making in courts. However, applications
of AI in medicine have been generally met with skepticism or overhype,
primarily due to the perceived black box nature of these computer algorithms.
We argue that physical principles can help uncover the mysteries of AI and
unlock its potentials. In this talk, we will discuss how physics and AI are inter-
related. We will draw examples from on our experiences in radiation oncology
and others for predictive analytics and decision making. We will further highlight
future AI implementation potentials and discuss current technical and ethical
challenges for delivering safer and better treatment for cancer patients using
physics empowered AI technologies.