Her ideal exam would not test memorization and replication but what students can apply, integrate and create in the real world. It might run over a few weeks: students analyze the assigned literature, form a problem-case scenario, and design a project or program to solve it — defining the target population, the problem and their reasoning, giving a step-by-step view of their thinking and creativity. AI could be used in the process, but students would note the hardest stage where its help was needed, then reflect on what they did well and would change next time, encouraging constant evolution. A defense of the idea would be included, helping students practice before they must defend a thesis.