If you are interested in new technologies, you have certainly heard of ChatGPT. This conversational artificial intelligence, developed by OpenAI, makes it possible to generate code, hierarchical plans or even entire texts from a simple request.
For an article titled“This article was written by ChatGPT”, we had for example asked ChatGPT to write an entire article, specifying the number of words and the subject of the article. The AI is even able to generate these texts in several languages: French, English, German, Russian or Japanese.
If this super-powerful tool makes it possible to achieve beautiful things, it can also make it possible to circumvent the rules. This is the case, for example, in schools, where we observe a resurgence of cheating with ChatGPT 3. Artificial intelligence is indeed capable of generating homework or complete dissertations on hyper-complex subjects.
Students only have to copy the instructions and the AI automatically generates a text. If the answer is considered incomplete, it is possible to write to him ” Keep on going “ to push the analysis further. All in just a few seconds. Cherry on the cake, ChatGPT writes unique content every time. A nightmare for teachers who have a hard time spotting cheating.
How teachers counter cheating with ChatGPT?
In recent months, the copies written by ChatGPT have multiplied, so the faculty has had to find new methods to circumvent the AI. Since anti-plagiarism software cannot detect homework done with ChatGPT, teachers fight back and offer less homework.
In place, they offer face-to-face homework, surprise orals or study corpora,indicates Radio France. The newspaper The duty says that some literature professors have even detected some flaws in the AI.
They found that ChatGPT “knew” certain works very well. They therefore began to offer works that the AI mastered less. To take it by default, it is also sufficient to offer texts released after December 2021, the database from which the AI draws stopping on this date (for the moment).
Above all, teachers must train to spot texts written by the AI. It is in any case the advice given by MIT (Massachusetts Institute of Technology) which explains that tomorrow we will be confronted daily with texts produced by AIs. Some will even be part of disinformation campaigns, so we will have to know how to spot them.
Some teachers therefore consider also note the style in the students’ homework. In effect, ChatGPT is good at generating texts clean on complex subjects, but the writing remains simplistic, with very common words and a predictable syntax. A bit like… a robot.
We would love to thank the author of this short article for this amazing material
