When Chilean university professor, Diego Martinez, asked his 50 students if they had used ChatGPT to help with an engineering assignment, he was surprised to find that every one of them had.
The popular AI chatbot can generate coherent prose, including essays, stories, summaries, legal text, and even poetry about virtually any subject in response to users' questions and is designed to mimic a human conversation.
"We just have to realise that there is a new student in the room that is helping everyone," said Martinez, associate professor of industrial engineering at Chile's Pontifical Catholic University of Valparaiso (PUCV).
"Students use it for everything, to brainstorm, to synthesise information to improve grammar," said Martínez.
From Chile and Colombia, university teachers and students are using OpenAI's chatbot on a regular basis as a research assistant, writer, editor, and even code programmer, which they say can boost learning and reduce teacher workload.
In Mexico, Rafael Mendoza, a computer science professor at the Autonomous University of Mexico State, was also caught off guard by ChatGPT's boom.
"My students caught me by surprise. They got to AI before me, and I became suspicious as they suddenly started delivering perfect assignments, with no grammatical errors and perfectly structured," said Mendoza.
At Brazil's University of Brasília, art student Maicon Costa has been using ChatGPT for months for research and to stave off his writer's block.
"When it started, professors were terrified of it, but now they're trying to get us to learn how to use it," Costa said.
Critics have warned the tool could lead to greater misinformation, cheating and plagiarism and called for stricter regulation of generative AI.
Martinez noted that while ChatGPT can help improve teaching and learning students need to approach it with a critical mind, while teachers should be left to decide if they want to introduce their own guidelines on its use.
"ChatGPT is a very good tool but it's also good at lying, making mistakes, which are called hallucinations, and you have to be an expert to catch them," said Martinez.
"And, I don't think the students are there yet," he said. Universities in Latin America are debating the benefits and pitfalls of AI in higher education and society.
In March Brazil's biggest state-run university, USP, hosted a seminar on ChatGPT, as have other institutions in the region.
In Mexico, universities like the Monterrey Institute of Technology have shared recommendations with teachers and students on the ethical use of chatbots, such as attributing text written by AI and double-checking the sources it provides.
"It's not a matter of right and wrong. We need students to be responsible when using AI," said Edward Bermudez, who is co-drafting AI guidelines for the Ibero-American University in Mexico City.
The main challenge facing universities is how to ensure ChatGPT's already ubiquitous use does not lead to more cheating.
Some academics have warned their students' grades will be cancelled if they are caught using ChatGPT, while others have changed the way they are assessing students, including fewer take-home assessments and more hand-written essays and exams in class and oral exams.
"I'm going to try not to set essays outside of class because this immediately invites the use of ChatGPT," said philosophy professor, Andres Paez, at Colombia's Los Andes University.
One colleague, he said, is showing students examples of essays generated by ChatGPT and asking them to improve on it.
Some professors, particularly in law, economics and business schools, are teaching students to effectively use the tool, ask it better questions, and to be critical.
Teachers say their students are concerned that generative AI tools will make some professions obsolete, such as entry-level coders, paralegals and data analysts, leading some to question the value of a degree.
Plagiarism concerns have led some institutions like Sciences Po, one of France's top universities, to ban its use in January, as did India's Bengaluru-based RV University.
In the United States and Australia, some high schools have banned the use of ChatGPT to prevent cheating.
"I think the consensus is that ... prohibition doesn't get us anywhere," said Páez, who is also a researcher at Los Andes Artificial Intelligence Research and Training Center.
The writers are from the Reuters news agency