Dealing with climate change is among the greatest problems we face today. We use scientific models to predict climate change. These models are derived from scientific principles but rely on approximations to make the prediction problem tractable. The physical basis of climate prediction was recognized by the 2021 Physics Nobel Prize that was awarded to two pioneering climate scientists. This talk will discuss the history and philosophy of climate prediction, explaining how some fundamental expectations of physics, such as replicability of results and quantification of uncertainty, do not directly carry over to climate science due the immense complexity of the system. Although we have a good understanding of average properties of climate change, there is deep uncertainty regarding the extreme aspects (the tail of the probability distribution). Looking to the future, we need to assess climate risk and make decisions about climate change while considering this deep uncertainty.