ChatGPT is a great tool, provided it is used wisely. In Lyon, a university professor was confronted with the dark side of this text-generating artificial intelligence, or rather that of his students. After giving them a homework assignment, the teacher noticed great similarities between several copies. The culprits finally confessed: 50% of his students entrusted part of the writing of the assignment to the lights of OpenAI.
Seven out of 14 students cheated
Stéphane Bonvallet tells the Progress his dismay in recovering the copies supposed to define “the main features of the medical approach to disability in Europe.”. “We found the same grammatical constructionsexplains the man. The reasoning was conducted in the same order, with the same qualities and the same faults. Finally, they were all illustrated by a personal example, relating to a grandmother or a grandfather”. Having interviewed one of his students, Stéphane Bonvallet obtained confirmation that 7 out of 14 students had used ChatGPT to write the assignment.
But how to react in such a situation? “Having no framework currently prohibiting this practice, I was forced to note them”explains the professor, who gave a score of 11.75 to all of these works written with the help of AI. “This practice poses a real problem of integration of knowledge, because students no longer need to carry out any research to compose”says the teacher to our colleagues.
ChatGPT raises ethical issues
The problem posed by ChatGPT and spotted in Lyon is probably not insignificant, and this bad practice could be more widespread than it seems. On the other side of the Atlantic, the New York school district has decided to crack down. The department in charge of education has indeed blocked access to the conversational agent of OpenAI on the entire network and computers of the schools of the city. A measure to protect students’ education and prevent some fraud, but which is unlikely to prevent its use for homework.
Moreover, most anti-plagiarism software seems incapable of detecting the generation of a text by an artificial intelligence. Some will therefore have to adapt to identify the frequency of use of words, or even the absence of typos, a fairly characteristic feature of the content produced by ChatGPT.