What a Professor’s Chart Teaches Parents About AI Cheating in College Exams
Last Sunday, I was folding my daughter Lila’s college laundry at my kitchen table while my sister Sarah, a biology professor at a nearby public university, stirred a second cup of coffee. Lila, a sophomore, was scrolling through her course schedule for next semester when Sarah pulled up a spreadsheet on her phone and slid it across the table. “I made this for my department meeting this week,” she said. “I think you’ll want to see it.”
The chart was simple enough: the x-axis marked every semester going back four years, before large public access to generative AI. The y-axis marked the percentage of take-home exam responses confirmed to be AI-generated. A sharp upward line climbed from 7% in spring 2022, to 19% in fall 2022, to 32% in spring 2023, landing at 41% last semester. I did a double take. Forty-one percent? Almost half of her 250-person intro bio class was turning in AI-written work for major exams?
Sarah nodded, and explained she didn’t rely only on sketchy AI detection tools. Her team of TAs cross-checked every flagged response against the student’s past in-class work and writing assignments. Nine out of 10 flagged submissions were definitely not written by the student. That’s when Lila looked up from her phone and said, “That tracks. Half the kids in my statistics midterm used it last month. It’s not like it’s hard to do. You just upload the question and wait 30 seconds.”
I asked her why. I said it’s not like most of these kids can’t answer the question themselves. She shrugged, wiped a crumb of granola bar off the table, and said, “Most of the people I know don’t cheat for fun. We have four midterms the same week, a group project due, and a work shift that can’t get rescheduled. It’s not that I can’t answer the question. It’s that I don’t have three hours to write a two-page answer to one question when I have three other exams to study for.”
I grew up with the idea that cheating is a clear moral failing. When I copied a friend’s algebra homework in 10th grade, my mom made me stay after school to confess to my teacher and rewrite the full set from scratch on my own. I carried that lesson with me for decades: if you cheat, you’re cutting corners, you’re lazy, you’re letting people down. But sitting at that kitchen table, looking at that line climbing steadily up Sarah’s chart, that old narrative didn’t fit anymore. This isn’t a handful of bad kids making bad choices. This is almost half of a large intro class. That’s a system problem, not a kid problem.
Sarah made the chart not to punish students, but to convince her department to change how they assess learning. For years, open-book take-home finals were popular because they cut down on proctoring time and classroom space. Now the department is voting next month to swap those high-stakes take-home finals for three 20-minute, un-timed in-person quizzes spread across the last month of the semester. Each quiz covers a small chunk of material, adds up to the same final grade weight, and it’s far less overwhelming for students who already have a full plate. It’s a small, concrete change that already cut AI use by half in the two intro sections that tried it this semester.
As a parent, I’ve had to make small concrete changes too. A year ago, if Lila had told me half her class was using AI, I would have launched into a lecture about integrity and how cheating would hurt her long-term. Now I know that kind of talk just makes kids shut down and stop telling you what’s really going on when they’re stressed enough to consider cutting a corner. The first swap I made was the question I ask. Instead of “you wouldn’t cheat, right?” I ask, “When was the last time you felt so swamped with work that you thought about cutting a corner?” It’s non-accusatory, and it gets her talking instead of closing off.
Another small thing I do now: at the start of every semester, we sit down for 15 minutes with all her syllabi and map every big exam, paper, and project due date. If three major deadlines fall the same week, I help her draft a short email to each professor asking for an early extension on one or two, before the crunch hits. Lila says most professors are willing to shift a due date if you ask early, but most kids don’t ask until the night before, when it’s too late to say yes.
After Sarah left that night, we ordered pizza, and Lila told me she used AI once last semester. She had to turn in a paper outline by 5pm, and the university library’s access to the 12 research articles she needed was down for emergency maintenance. She used AI to pull the main arguments from the public abstracts she could find, so she could hit her deadline. Once the library came back online, she read all the full articles and rewrote her whole outline from scratch. She told her professor about it after, and the professor said that was exactly what AI should be used for. I didn’t yell. I just said, “That makes sense. I would have done the same thing.”
I still worry about it. I worry that kids will get too used to letting AI do the hard work of thinking through a messy problem, and won’t build those skills when they need them. I worry about the kids who don’t cheat, who spend all that time working hard, only to get the same grade as someone who finished their exam in five minutes with AI. That’s unfair, and there’s no easy fix for that. Sarah’s chart didn’t give us a clear solution. It just exposed how big the problem really is, and that it won’t go away if we just pretend it’s a handful of bad kids making bad choices. We’re all still figuring this out, parents, teachers, and kids alike. The only thing that’s clear right now is that shaming kids isn’t working. Changing the system one exam at a time is a good place to start.
Please indicate: Thinking In Educating » What a Professor’s Chart Teaches Parents About AI Cheating in College Exams