Where Objectives Come From: Why AI Needs Normative Modeling
A loss function is only one of the four parts of a decision problem. From Wald’s decision theory to normative modeling, this post asks where objectives come from, and why modern AI needs to treat them as hypotheses to be tested.