Porte-Plume

Partager les belles expériences à l’École

Retour

AMC Blog - Lycée Mona Ozouf de Ploërmel

Things to look out for when using AI

Publié le 12/10/2026

parShi__d ,
Photo de Steve A Johnson sur Unsplash
Photo de Steve A Johnson sur Unsplash

Nowadays, a lot of students use generative AI for their homework, their assignments, or other things they may need help for. However, a lot of us use it wrong. That raises the question of how we can notice these wrong ways of using genAI and how to avoid them.

What is Generative AI ?

To start, generative AI is defined by all AI that generates text, images, videos, or other media in order to respond to a prompt given by the user. The most common example is AI Chatbots, such as ChatGPT, Gemini, Midjourney, Claude and a lot more others that have risen up. That’s why most students use it for their work : instead of doing it, they can simply type in the question on a genAI app and get the answer or an entire essay written in an instant ! Isn’t that so much easier to do ? That’s the first trap most of them fall into.

Laziness: Not Learning

School is a place where you’re supposed to learn. A part of it is done with homework, that can let you exercise yourself outside school. While I myself have opinions on whether homework is something that should be done or not, I do admit that using genAI for it is not great for you. By using genAI to do YOUR work, you completely remove the working aspect of homework. Because a program did it for you, you don’t actually learn anything : you don’t make any mistakes that you can grow from, and it incentivises you to be so incredibly lazy.

Of course, there are ways to use genAI to help you learn, like by getting it to explain something you might’ve misunderstood from your teacher. But you shouldn’t take everything it says as fact. It often gets things wrong.

Misinformation, Bias...

GenAI gets its information from the internet. Because of that, its sources can get... questionnable. For example, it may gather information from a biased site, or sites that are straight up spreading misinformation. And that’s why it doesn’t always say everything right : around 51% of answers to prompts given by ChatGPT were wrong, incomplete or imprecise, according to research done by BBC in 2025. That means in over half of cases, it’s not fully reliable. And yet, a lot of people see what it says as the objective truth, when that isn’t the case at all. It’s also quite easy to fix this problem. All you need to do is specify in your prompt the use of a specific source that you know you can trust to not have misinformation or bias.

Some sources that you could use are things like news sites, including BBC or The Guardian. But don’t be so hasty – maybe your source doesn’t want to be used by genAI.

Theft

In order to get an AI model up and running, a critical step is to train that model. Oftentimes, that training is done by stealing content from other people, whether it’s art, articles, research, genAI doesn’t produce anything original. All it does is take an input and use that to give an output. Of course, a lot of artists won’t take too kindly to their art being stolen like that. That’s the most obvious example, but a less obvious example is something that happened not too long ago in the field of mathematics. Two mathematicians were working on a problem – the Navier-Stokes equations, and were getting close to solving it. It was one of few mathematical equations that hadn’t been solved for nearly a century, and had a cash prize of 1 million dollars for whoever could find the answer. These mathematicians were using AI tools in order to help them, which was a mistake, as soon, OpenAI, the company behind ChatGPT, had caught wind of it. They used 10 000 differents chats with over 3 million messages in total to solve the problem using the mathematicians’ work so far in just 88 hours. The proof hasn’t been reviewed yet, but this situation raises the question of who deserves the credit ? The mathematicians or the AI ? And who does the prize money go to ?

To solve this problem, you can see if the author of whatever page or image you’re working with consents to AI being used on their work. And most importantly, you need to ask your teachers if they’re fine with your use of AI.

Conclusion: My Advice

To conclude, when using AI for schoolwork, you need to watch out for these three things :

  1. Wording prompts so that the AI doesn’t do the work for you, but teaches you how to do it
  2. Direct it to use sources that are known for being trustworthy and objective
  3. Make sure whatever source is used doesn’t mind the AI using their work as a source

If you don’t do these 3 things, then, in my opinion, you’re not using genAI correctly.

Recently, I found something that you can do with genAI that ticks all 3 boxes to help you learn :
You use AI to generate a podcast, a song, etc. talking about your notes.
This solves 1 because it’s giving you a new medium to learn with, it doesn’t do the work you’re supposed to.
It also solves 2 by making the source your own notes because if it gets something wrong, then it’s you have no one to blame but yourself.
Finally, it solves 3 because the source is your own notes, therefore, if you want to give AI those, that’s your own choice, not anyone else’s.
While I personally haven’t tried this method, on paper, it seems like it could be worth a try, especially for students who struggle to study with traditional methods. However, do be careful when doing this – make sure to review the result of the generation so the AI doesn’t get any details wrong and to correct it when it does.

Personally, I won’t (and don’t) use AI. While I am aware of how it’s possible to use it as a tool, I still have strong Anti-AI beliefs, so the only times I could use AI would be if I am truly desperate.

Partager le billet