The Case Against Optimizing Our Education to Death

Last semester, I took Calculus III. I submitted my first problem set without using AI, got a 57 percent, and was flummoxed at the class average of 97 percent. I asked a friend in the class for help, and he responded incredulously: “You didn’t run it through Chat?”

On February 11, I attended a dinner-table discussion sponsored by the Undergraduate Community Initiative and the CC-SEAS Integrity Advisory Board. The theme of the dinner was generative AI: How, why, and by whom it was being used. A good mix of students were in attendance, as well as Dr. Victoria Malaney-Brown and Dean Jonathon Kahn, director of academic integrity and dean of community and culture, respectively. As the goal of the discussion was to talk openly about the subject, I admitted that I had succumbed to the pressure. It was like a dam had broken; around the table, everyone admitted to using AI in their more technically-oriented classes, if not all of them.

My own use of AI had started as a way to catch up to the crowd of already-users—it wasn’t ill-intentioned, and it certainly did not begin as a substitute for doing my own work. After the Calc III incident, I ran all my problem sets through AI before submitting them. The line between using it to catch mistakes and using it to solve difficult problems soon blurred: The nature of AI was such that it quickly made me feel incompetent—it was just so fast, so good at spitting out exactly the right solution, given the right prodding. Eventually, I was using it as a crutch for all the problems I didn’t understand, rather than actually doing the hard work of untangling the solution myself.