Clear guidelines needed on AI use in classrooms: experts

Editor in Chief of Android Central Shruti Shekar explains the dangers of utilizing ChatGPT as a mental health resource and discusses safety tips for parents.

Educational institutions are looking at developing ways to handle the increased use of artificial intelligence programs and large language models (LLM) among students.

A 2024 survey from KPMG found 59 per cent of students in high school and college surveyed used generative AI in their schoolwork, a 13 per cent increase from the previous year. The use was primarily to generate ideas (46 per cent), research (41 per cent) and write essays (32 per cent).

Michael Holden, an assistant professor in the University of Winnipeg’s faculty of education, said AI has been a dominant topic of conversation among teachers since ChatGPT became publicly available in 2022.

“The initial reaction from a lot of schools and teachers was just sort of shock and awe,” he said. “Now nearly three years into it, I think really where we’re seeing a lot of the conversation go is we really need clear guidelines, clear policies, and clear support for teachers, parents and for students, so that we have an understanding of what this space is that we’re playing in, and how we are all moving forward on this work.

“If you don’t have those clear guidelines, it’s really hard to have a united front on how you’re actually going to support student learning.”

At the school division level, Christian Michalik, superintendent of Winnipeg’s Louis Riel School Division (LRSD), said the division is developing guidelines for staff and students to address the presence of AI programs such as ChatGPT.

“We are well aware that artificial intelligence is rapidly transforming our lives and the way we work and learn,” he said. “In LRSD, we see both opportunity and challenges, and our focus is to have students and educators learn to use AI ethically and critically as well as creatively.”

‘A dishonest element’

Since AI became dominant, multiple concerns have been raised, including academic dishonesty, incorrect information being generated, privacy and security concerns, and bias in algorithms.

“People are unduly relying on the technology to produce their work product,” said Katie Szilagyi, a professor at the University of Manitoba. “So, there’s a dishonesty element if someone is representing AI-generated work as their own.”

Szilagyi added the purpose of education is to learn, and using programs can defeat the purpose.

“If you start outsourcing that kind of academic output to an LLM, there are concerns about whether people are really learning the materials, whether they’re examining or being tested on the materials accurately, and if you’re going to be giving someone a credential, did they really earn that credential If they are unduly relying on AI?” she said.

Holden has said the rise of AI has seen teachers change how they’re assessing students.

“You don’t want students offloading that work to the AI tool,” he said. “If you care about a student’s ability to write a really persuasive argument, you need to get them to do some writing.”

Michalik said these are concerns that will need to be addressed with collaboration between schools, school divisions and the province.

“The more we come together to talk about concerns and to collaborate, the more likely we are to see less worry around that concern,” he said.

Responsible use needed

RRC Polytech recently launched a micro-credential focused on AI in the classroom. Carleigh Friesen, chair of adult education and teaching excellence at the college, said the presence of generative AI and LLMs is similar to how disruptive the calculator was in education.

“When it started getting introduced into schools, we were told that students are going to stop thinking, and we’re seeing the same kind of disruption,” she said.

Friesen said she believes generative AI should be seen as a possible tool in the toolbox for teachers and students.

“We don’t believe that banning is the right approach,” she said. “It is kind of a head-in-the-sand approach. We believe that the toothpaste is out of the tube, and we have to find what are the ways that we can use this tool responsibly. How can it augment critical thinking with our students? How can it support and build skills for our student body?”

Holden said the way the discussion on AI is framed, such as outright banning it or accepting the inevitable, is a mistake.

“I don’t think those are helpful conversations,” he said. “I think the most important thing is to emphasize that idea of how we are going to keep the focus on the learning for the students that we have in the classroom right now.”

Szilagyi, who has helped the University of Manitoba develop its own policy on appropriate AI use, said automation bias and complacency remain a concern for people using AI.

“There’s a bias in that you think that the AI is better than you at the task, and there’s a complacency where you’re going to let them take care of it,” she said.

“In instances when we interact with automation, we know that people are lulled into these psychological states where they can be biased and complacent and allow the automation to take the driver’s seat. That’s one of the things that I really worry about,” she said.

She added the program is not really communicating with people but rather simulating and predicting what the next word should be in a text.

“At the heart of human communication is the ability to convey meaning, and at least at this time, that’s not something that the AI is capable of doing,” she said.