There is no such thing as 'alignment'. I want the model to be aligned with my intent, which can be whatever I want it to be. Your intent is not a more real or objective intent than mine.
More specifically, alignment is a reference; 'alignment with X'. A reference necessitates a referent.
You can't reify 'alignment', it is not an intrinsic property of a model.
More specifically, alignment is a reference; 'alignment with X'. A reference necessitates a referent.
You can't reify 'alignment', it is not an intrinsic property of a model.