�첥��Ƶ

When Robin Sutherland-Harris saw generative artificial intelligence (GenAI) emerging in the autumn of 2022, she knew her role as an educational developer at �첥��Ƶ's Teaching Commons was about to expand.

Her job is to support faculty with pedagogical advancement, and with her academic background in researching large language models and natural language processing, she was a natural match for helping teachers integrate GenAI into their classrooms.

“There were some apprehensions around academic integrity and things like that,” says Sutherland-Harris. “I was really eager to start framing the conversation as one that had deep pedagogical considerations, that it wasn't just about enforcing academic integrity or about a new technology, but it was more complex.”

She thinks no one fully comprehended some of the challenges AI would present, including difficulties faculty would face in preparing students to use AI effectively.

“I really wanted to make sure we were having some space for a pedagogical conversation,” she says. “I just started doing things with faculty where I could, to make spaces for those discussions. And now it is almost the entirety of my portfolio at the Teaching Commons.”

Sutherland-Harris now teaches a course for faculty called “AI and Education,” addressing challenges such as policy-making, ethical use and integrating AI into teaching. The course can count toward a Certificate of Proficiency in Teaching for eLearning. She also co-leads York’s AI Community of Practice, and has seen it grow to more than 200 members, which she believes is a reflection of the growing interest in AI education.

According to Sutherland-Harris, business professors are more likely to embrace this new technology because of their students’ interest in the private sector, where GenAI is being experimented with the most. An excellent example of that, she says, is Andrew Sarta’s class assignment presented at a recent conference.

Sarta is an assistant professor in the School of Administrative Studies in York’s Faculty of Liberal Arts & Professional Studies (LA&PS). In his first use of AI in his teaching, Sarta combined AI and virtual reality (VR) in a course at York’s new Markham Campus that focuses on training students in creativity and innovation.

Sarta’s second-year students were asked to work in groups and choose one of the 17 United Nations Sustainable Development Goals, then develop a business venture that could address their chosen goal. Examples included achieving gender equality, reducing poverty and building sustainable cities. The students used VR headsets to meander through the Oculus Wander app in pre-designated locations where they might find inspiration to solve their problem. They then used ChatGPT, the GenAI chatbot, for supplemental research.

“If you're sitting at your desk all day, if you're talking to the same people all the time, it's difficult to understand what other people's problems are and how you might solve those problems,” Sarta explains. “The idea behind VR was to mimic that experience, to get the students out of their space to explore different environments where they might be inspired by something.”

The VR portion of the assignment was a hit. Students were excited about their immersive VR experiences and requested more of them and more dynamic environments to better understand local issues. Sarta plans to make the VR experience even more interactive in future.

Surprisingly, Sarta’s students found their own intuition and traditional research methods more effective than ChatGPT for problem identification. Going forward, he will teach students how to more effectively use large language models.

Heather Lynn Garrett, a course director and contract faculty member in the LA&PS Department of Sociology, is also experimenting with GenAI. She does not allow students to use AI in her classes, but she has used ChatGPT in her second-year research methods course for a unit on qualitative data analysis.

Garrett explains the process: she used ChatGPT to generate coding categories from interview transcripts. Students then generated themes from transcripts, coded them manually, then compared their results with the AI-generated codes.

And in the ultimate question of man versus machine, man won this round. Despite ChatGPT’s initial success in producing 29 categories, it struggled to reduce them to six and often hallucinated content, meaning it made up the answer. The students’ manual work was far superior.

“I came up with the exercise because I want my students to be critical producers and consumers,” Garrett says. “When they did their reflections on it, they were happy to have had the opportunity to compare ChatGPT to what they could do. If I had just said, ‘Use ChatGPT to do it,’ they wouldn't have had the opportunity to have tried to do it themselves first.”

Garrett says she plans to continue exploring and integrating AI into her teaching, always balancing its benefits with the need for critical thinking.

Sutherland-Harris says it’s no surprise that teachers can’t just introduce something with GenAI into their classroom with the assumption students will immediately understand it or see its value.

“It's a thing that we all have to learn about and learn how to use together – which is, I think, a really rich pedagogical space.”

With files from Julie Carl

Tags: