Is There a Subjective Versus Objective Goal in Machine Learning?

I think the initial framing of the problem for me was always about, what's the correct objective? When started working on this problem, i definitely did not have access to the sentences i just said. My background comes from a planning, rebotic context. So to me, most a applications can be thought of as some version of a markof decision process. They're very clearly delineated into objective perties of the world are properties of a state transition matrix conditioned on your actions. The reward function provides the subjective information, the normative information that tells you andf right from wrong.