My College Student Is Required To Use AI For Programming: What I Learned As A Parent
The iced tea my husband spilled was still seeping into the tablecloth when our son Leo dropped the bomb at spring break dinner. “My advanced programming prof is requiring us to use AI for our midterm project,” he said, stabbing a piece of cold sesame chicken. I froze mid-bite. I’d spent the last two years reading headlines about AI cheating in colleges, about kids getting expelled for using bots to write essays and code. The last thing I expected was a professor requiring it.
I said what any skeptical parent would: “Wait, that’s allowed? What are you even going to learn if it writes all the code for you?” Leo rolled his eyes, the way 20-year-olds do when their parents are obviously out of touch, and pulled his laptop out of his backpack, balancing it on the edge of the table between the takeout containers and the salt shaker. “It’s not that simple,” he said. “Let me show you.”
The assignment prompt was nothing like I imagined. Leo’s class was tasked with building a custom inventory app for a local small business or non-profit. The requirement wasn’t “let AI do your work” — it was “generate a first draft of your code using any AI tool of your choice, then debug, refactor, and adapt that code to meet your client’s specific needs.” Grading was broken down: 10% for the AI’s initial output, 70% for Leo’s changes, and 20% for a written breakdown of every error the AI made and why his fix worked better.
Leo’s client was a neighborhood animal shelter that needed to track medication expiration dates for foster dogs. The AI he used spat out 300 lines of generic inventory code in 12 seconds, but it missed almost every one of the shelter’s specific needs. It didn’t include a field for expiration dates at all. Its sort function crashed when the list of foster dogs hit 25 entries. It didn’t flag duplicate names, which is a common problem at a busy shelter. Leo spent 12 hours rewriting the error handling, adding the required fields, and testing edge cases — like when a medication expires the same week as a big adoption event.
“So you’re basically just cleaning up AI’s mess? That’s the homework?” I asked. He leaned back and shook his head. “That’s what I do at my internship, Mom. All the senior devs use GitHub Copilot to spit out boilerplate code. No one writes 100 lines of basic sorting from scratch anymore when a bot can do it in 10 seconds. The hard part isn’t writing the basic code — it’s knowing what the bot got wrong, and how to fix it for what we actually need.”
That conversation stuck with me. For years, I’d bought into the idea that any AI use in school is cheating. But this approach flipped that script on its head. Leo told me that last semester, before the prof changed the policy, more than half the class was using AI secretly to get assignments done faster. No one talked about it, no one learned how to check the output — they just turned in the code the AI gave them, and half the time it didn’t work, and they couldn’t figure out why. Now, the secret is out, and the skill of checking and editing AI output is the whole point of the assignment.
I’ve talked to a few other parents since that dinner, and most of them have the same reaction I did: immediate panic. If your kid comes home and says their programming class requires AI, here’s what I’ve found actually helps, instead of spiraling. First, ask to see the full assignment prompt before you say anything. I almost gave Leo a 20-minute lecture about learning to do things the hard way before I saw that the hard work was still there — it was just a different hard work. Second, even if you don’t know a single thing about programming, ask them to show you one mistake the AI made. You don’t have to understand the answer, but asking the question reminds them that their job isn’t to copy-paste AI output — it’s to check it. When I asked Leo that question, he talked for 15 minutes about how AI forgets small, specific requirements all the time, because it’s trained on generic code. That told me more than any syllabus could about what he was actually learning.
Third, if you’re a teacher looking for a way to handle AI in programming classes, this model works way better than banning it. Banning AI just pushes it underground, and kids don’t learn to use it responsibly. Requiring it, and grading their ability to critique and improve it, turns a cheating risk into a job skill that aligns with what they’ll actually do after graduation.
I still have questions, of course. I wonder what will happen when AI gets so good it rarely makes those obvious mistakes — what will professors teach then? I worry that kids who are just starting to learn programming don’t get enough practice writing basic code from scratch to understand how it works. Leo says that’s already addressed: in his intro 101 class, you still have to write all your own code for the first six weeks, to learn the basics. Once you know what a loop or a function actually is, then you learn how to use AI to do the boring stuff.
Last week, Leo sent me a photo of the shelter volunteers testing his app, all of them leaning in to add their foster dogs’ medication dates without crashing the system. He got an A on the project, but the thing he kept talking about was how much easier it was to focus on the part of the project that actually matters — helping the shelter — instead of wasting time writing lines of code he could get a bot to spit out. It’s not the way I learned to code, that’s for sure. But it’s also not the end of learning that I was afraid it would be. We’re still figuring it out, same as every other generation did when new tools changed how we work.
Please indicate: Thinking In Educating » My College Student Is Required To Use AI For Programming: What I Learned As A Parent