Audio summary of the essay explaining the representation–motivation gap in artificial intelligence. The framework proposes a categorical model in which alignment requires preserving normative structure across mappings from human values to machine representations and agent actions.