Romain Rouillard 9:00 p.m., January 26, 2023

Sciences Po has called on its teachers to formally ban the use of ChatGPT, this artificial intelligence capable of writing texts and recording mountains of information.

Students who circumvent the rule could incur permanent expulsion from the establishment.

If the ChatGPT robot amazes the scientific community, it is greeted with more apprehension by the teaching staff.

Many students might be tempted to use this artificial intelligence – capable of writing text from a simple question – to compose their university work.

The machine has swallowed all the texts available on the Internet until 2021 and presents a fairly impressive level of knowledge on a multitude of diverse and varied subjects.

"This tool (...) strongly questions education and research actors around the world on the subject of fraud in general, and plagiarism in particular", recognizes Sciences Po in a letter addressed to teachers and made public by BFMTV.

>>

READ ALSO

- ChatGPT: why the conversational robot remains largely perfectible?

Faced with the magnitude of the phenomenon, the famous school decided to take the bull by the horns by indicating in this letter that the use of ChatGPT is "strictly prohibited when producing written and oral work".

Offenders are exposed to severe sanctions "which may go as far as exclusion from the establishment or even from higher education".

Sergueï Guriev, director of training and research, and author of this letter, calls on the teaching team to be vigilant: "I know I can count on your commitment to guarantee quality learning and diplomas based on rigorous academic integrity" . 

Difficult to prove use 

However, the establishment does not intend to bury the subject under the carpet.

"The issues related to these language generation tools will inevitably and rapidly change teaching practices and teaching evaluations", writes Sergei Guriev.

And to announce the forthcoming holding of a conference on the teaching and research of the future "in an ecosystem where AI is taking an increasingly important place". 

If ChatGPT causes a lot of torment to the teaching staff, its use remains extremely difficult to prove.

This week, a professor of sociolinguistics at the University of Aix-Marseille found himself confronted with the problem.

"There you go: first big hint of an assignment written using ChatGPT. No apparent copy-paste, but a text written in an impeccable and quite subtle way, without mistakes, from a student who had had 4 /20 and 7/20 on his first homework. What am I doing?" he tweeted. 

Here it is: first big hint of an assignment written using #ChatGPT...


No apparent copy-paste, but a text written in an impeccable and quite subtle way, without mistakes, from a student who had got 4/20 and 7/20 on his first assignments.


What am I doing?...

– Mederic Gasquet-Cyrus (@MedericGC) January 25, 2023