Given some experimental data, we often want to estimate the unknown parameters behind it. Sometimes however, all we really want to know is what’s going to happen next? In this talk, we follow the 2013 SIAM review paper by Chad Lieberman and Karen Willcox and look at how the inference-for-prediction method incorporates the end goal into the inference process, sacrificing parameter estimation accuracy for online efficiency. We shall investigate the motivation behind Goal-Oriented Inference and the control theoretic inspiration of Balanced Truncation, before looking at both a traditional method for these parameter dependent problems and the inference-for-prediction method.