Contending Frames: Evaluating Rhetorical Dynamics in AI