Delegating intellectual work to AI is, at each decision point, fully rational. Yet the cumulative effect is the structural erosion of critical thinking capacity—with no rational basis for halting the delegation. This paper analyzes this paradox, termed rational surrender, through four mechanisms. First, cognitive delegation is self-reinforcing at an accelerating rate (§2). Second, the psychological motivation to resist delegation—the pleasure intrinsic to discovery, construction, and dialogue—is simultaneously eroded (§3). Third, under this dual pressure, critical thinking education becomes a foregone experiment: an intervention whose outcome is predetermined by the asymmetry of speed and scope between AI advancement and educational reform (§4). The standard prescription to “use AI critically” harbors a self-defeating circularity: for novices, delegation precedes critical thinking; for experts, critical thinking itself rationalizes delegation (§5). Structural interventions can exploit the time lag between lower-order and higher-order cognitive delegation, but these are palliative, not structural solutions (§6). The paper concludes by tracing the terminal trajectory of intellectual agency beyond education (§7). This analysis was itself conducted with AI assistance, instantiating the structure it describes.