Because exams usually test how you reason, not just whether you can reach an answer. AI can help with recall or surface explanations, but if the task requires applying concepts under constraints, it won’t replace actual understanding. If it does help too much, that probably says more about the exam design than the tool.