AI can do your assignment, but are you actually learning?

I remember the first time I used AI at university. It was during the spring semester of 2022. In that semester, we had this group project where we had to build a travel website. The idea was to create a platform where clients could book activities and tours for their perfect trip. We had to build the website, set up the database, and get everything working together.

I’ll admit, I’ve never been a great coder. Sure, I passed the exam, but it took me a lot longer than some of my classmates. And for this group project, there was quite a lot to figure out, and I remember finding AI genuinely useful. It was my first time interacting with this new technology and it helped me work through problems and get things done that otherwise would’ve taken me much longer.

But then the following semester, I had less coding classes and more classes that required handing in essays or presentations. Most of my work was focused on learning other subjects, writing and explaining concepts. And this is where I started noticing a difference. I found AI pretty terrible at writing essays that actually said what I wanted to say.

I’d ask it to help me write something, get a wall of text back, and then spend ages reading through it, correcting it and trying to make it sound like what I had originally wanted to express. At some point, I started wondering whether it would’ve just been faster to write the thing myself in the first place. AI can be incredibly useful, but it really depends on what you’re using it for and, perhaps more importantly, how well you understand the subject yourself.

I noticed something similar later when I became a student assistant for a course at university. I could see students using AI to create models for their assignments. And having worked through the same material myself, I knew how time-consuming it could be to fix those models when they didn’t actually do what the assignment required.

The AI wasn’t necessarily the main problem. It was that the student didn’t fully understand what they were asking it to create in the first place. And how would they know whether the model was correct? If you don’t understand the subject, you might not even recognize that something is wrong. You can end up spending hours trying to fix an output you never really knew how to evaluate in the first place.

Of course, if you’ve spent time learning the subject and understand what you’re doing, AI can be much more useful. You can give it better instructions, recognize mistakes and make informed decisions about what to change. You become better at using the tool because you actually know your stuff. But that requires doing the learning first, or at least, not skipping it entirely.

And I think this matters particularly at university, because the whole point is to develop the knowledge and skills you’ll eventually need to use on your own. AI might help you prepare an assignment, but it’s you who has to sit the exam. You can’t bring the AI in with you.

Which makes me wonder whether students are sometimes confusing getting an assignment finished faster with actually learning something faster. Because those two things aren’t really the same. Maybe AI does save you time, but sometimes it just moves the work somewhere else. Instead of spending time figuring out the problem yourself, you’re spending time correcting an answer you don’t fully understand. And if you rely on it too much, you might finish the assignment without ever developing the skills the assignment was supposed to teach you.

I’m not saying students shouldn’t use AI. I used it myself, and I can see how useful it can be. But I think there’s a difference between using AI to help you learn something and using it to avoid having to learn it. And at university, that difference matters. Because in the end, it’s still you who has to understand the subject. It’s you who has to perform. And if the AI disappears tomorrow, or you’re sitting in an exam without access to it, what are you actually left with?