(Paris) Feared as potential tools for cheating or plagiarism, ChatGPT and other artificial intelligences are now banned from schools and universities all over the world, a “shortsighted” reaction for its defenders.
Ever since ChatGPT and its automatically generated texts became publicly available in November, concerned schools have been trying to stop their students from using it, both during exams and for homework.
Elon Musk, one of the founders of OpenAI, the start-up that created ChatGPT, triumphantly tweeted in early January: “It’s a new world. Goodbye homework! “.
The first major European university to try to block, Sciences-Po Paris at the end of January prohibited its students from using ChatGPT for any written or oral production, under penalty of exclusion.
The French Minister of Education even mentioned more global measures. “We will have to intervene on this, we are thinking about the right way to intervene,” said Pap Ndiaye Thursday on France Inter. “It is clear that if not we have an adversary in the matter, but in any case we have to integrate these new data into the work of pupils and teachers”.
The texts produced by the AIs are however, according to him, “quite different from what the students are able to write and the teachers are able to see the difference”.
Paper and pencil
In four of Australia’s six states, inside public schools, ChatGPT was banned from use in January thanks to a firewall, and the app banned on mobile phones for students on site.
The most prestigious universities in the country, like American universities, plan to increase on-site exams “with paper and pencil” or monitor screens for distance students.
New York City has similarly banned ChatGPT in its public schools on all devices, citing a lack of “critical thinking” and fear of “plagiarism”. Schools in Seattle and Los Angeles have followed suit.
In India, RV University Bangalore has banned it on its campus and scheduled more snap exams.
In Great Britain, the Office of Examination Regulation wants to set up a charter for schools. A Member of Parliament recently caused a stir in December by reading with a speech written by ChatGPT “in the manner of Churchill”.
At the University of Strasbourg, in France, around twenty students, who had cheated using this tool during a distance exam, then had to take it again face-to-face.
Image platforms like Getty Images and Shutterstock have banned images created by AIs like DALL-E, Midjourney and Stable Diffusion. Computer code forum Stack Overflow has banned posts produced by ChatGPT, saying they contain too many errors.
Researchers are also asked to abstain. At the beginning of February, the American scientific journals Science and Nature warned that they would no longer accept ChatGPT as an author and asked researchers who use it to mention it.
Calculator
The International Conference on Machine Learning, held in January in the United States, refused presentations made by ChatGPT, unless it was the subject of the study.
In front of these beginnings of sling, OpenAI has just announced a program which helps to distinguish a text written by ChatGPT from a text written by a human, but for the moment, he admits himself, “not entirely reliable”.
Proponents of the tool, such as Sébastien Bubeck, Machine Learning researcher at Microsoft, protested against “short-sighted” reactions. “ChatGPT is part of the future, banning is not the solution,” he tweeted.
Many critics of the ban point to the measures eventually dropped against pocket calculators or Wikipedia at school. Academics advocate framing rather than prohibiting, such as the vice-rector of the University of Neuchâtel Martin Hilpert, who prefers a discussion after submitting an assignment. “We will see quite clearly if the student has mastered his subject,” he told the Swiss press.
Bernardino Leon, researcher at Sciences-Po, also pleaded for AI in a column in Le Monde dated Friday, which according to him can help creativity.
“When calculators are used in education, students’ operational and problem-solving skills improve,” he argued.