Feb 17, 2021
In machine learning, typically optimization is done to produce a model that performs well according to some metric. Today's episode features Evan Hubinger talking about what happens when the learned model itself is doing optimization in order to perform well, how the goals of the learned model could differ from the...