Brown University economics students who had access to AI tools during a take-home midterm scored higher than those without, raising a question that stretches far beyond grade distribution: what does it mean to learn when the tool doing the intellectual heavy lifting is no longer you? The temptation to use AI for academic work has become almost frictionless. A student can paste a problem into ChatGPT and get a solution in seconds, complete with explanation. The immediate result looks like mastery. The long-term question is whether anything has actually been learned. Einstein, It argues, thrived on struggle, on wrestling with problems until the answer emerged from his own thinking. That cognitive wrestling is what builds understanding. When AI solves the problem first, the student never gets there. The dilemma isn't unique to Brown, and it's not going away. Schools are now caught between preventing cheating and acknowledging that AI is a tool their students will use professionally. But there's a real difference between using AI as a tutor or collaborator and using it to replace the thinking itself.