Back in 2017, a collection of academic professionals assembled at the University of Manitoba to share strategies and resources to promote learning experiences centred on integrity for post-secondary learners across the province. Little did they know how much value this event would bring to educators and learners in post-secondary institutions throughout Manitoba.

Academic Integrity Inter-Institutional Meeting (AIIIM) 2024

Held on May 8, 2024, at Assiniboine College’s (Assiniboine) Portage/Southport Campus, AIIIM 2024 was attended by students, instructors, librarians, educational developers, administrators, and staff from post-secondary institutions across the province.

“We were proud to be a sponsor and attendee at this engaging, insightful, and informative annual event,” said Carley McDougall, executive director of Campus Manitoba. “Connecting with the participants and observing how engaged the learning community is around academic integrity was rewarding, and seeing them embrace ways to the best possible learning experience for students was truly uplifting.”

“We chose Portage La Prairie for the event this year for a few reasons,” shared Josh Seeland, the manager of library services at Assiniboine College and event chair. “It’s a little more than an hour away from Winnipeg, and we wanted to introduce everyone to a post-secondary presence that they may not have experienced without the event.”

Through the ongoing support of event sponsors, Campus Manitoba, Manitoba Flexible Learning Hub, Brandon University, Assiniboine College, and the University of Manitoba, there was no charge for participants to attend the event. Keynote speaker Dr. Susan Bens from the University of Saskatchewan addressed the audience with “5 Things I’ve Learned About Academic Integrity,” followed by a workshop led by Dr. Brenda M. Stoesz of the University of Manitoba, Trevor Simpson from Assiniboine College, and Josh Seeland: “Using Generative Artificial Intelligence with Integrity in Teaching and Learning.”

Following a delicious lunch catered by Burnside BBQ, Dr. Lincoln Gomes from Navitas Canada and the International College of Manitoba and Dr. Claudius Soodeen from Red River Polytechnic facilitated a workshop around “Policy and Practice: Interpreting Academic Integrity.”

Dr. Brenda M. Stoesz led the final workshop of the day, which explored “The Future of Academic Integrity in Manitoba.”

Learning with Artificial Intelligence

During last year’s Canadian Symposium on Academic Integrity 2023, held at the University of Manitoba, there was a lot of unknown around generative artificial intelligence (AI) and large language models. At AIIIM this year, the focus was more on what can be done with AI in teaching and learning instead of dwelling on its negative impact on the validity of assessment. The workshops included many hands-on activities for everyone to experience first-hand what can be done with the available tools.

“It was interesting to see where institutions were at in their understanding and use of generative AI,” said Josh Seeland. “Where some weren’t authorized to do much with it yet, depending on their institution’s policies or procedures, others were on their way to becoming advanced practitioners, building experience by implementing different approaches.”

Privacy Pitfalls

“We often hear about free sites or commercial software that claim to detect artificial intelligence use,” shared Josh. “And while some of these tools are quite good, we’re seeing reports and studies that show they’re really unreliable when there’s any mix of human and AI. Some of them are good when the written content is 100% AI or 100% human, but for any combination of the two, the number gets skewed consistently across these detection tools.”

However, using AI detection tools to determine whether an assignment was completed by — or with — AI introduces new challenges. Data security remains one of the biggest considerations for all of these tools. In Stoesz’s workshop, she explained that since the students’ papers are their intellectual property, feeding them into the AI-checker potentially violates their rights, regardless of how accurate the checker is — or isn’t. These verification tools tend to look for specific patterns and structure, but academic writing is inherently structured and formulaic, so even if it was 100% written by the student, it might be flagged as being partially (or even wholly) written by AI.

Critical Evaluation

A potential solution to eliminate the concern over AI-generated submissions is to evaluate the process, not just the final submission. Kristy Lacroix, special projects coordinator at Campus Manitoba, explained, “In the workplace, it’s rare to present the final project and call it a day. There is an expectation that we, as professionals, always be prepared to discuss our processes and how we reach our conclusions. I believe AI is the future, and it can help everyone save time and effort by providing a solid starting point for a wide range of tasks and projects, but it’s still simply a tool that everyone — faculty, staff, and students — need to learn how to wield safely and effectively. Generative AI is not a replacement for true understanding or knowledge, but it is a powerful tool that can enhance our abilities.”

“We need to find ways to leverage these AI tools to create a better learning outcome,” explained Josh. “If paraphrasing or synthesizing ideas aren’t the learning outcomes, maybe it’s okay to use AI. Students need to know how to use it to do research, outlines, or an annotated bibliography, but rather than having them hand something in, they can either defend their submission with a viva voce or something like a thesis or dissertation defence. It’s not common in community colleges or undergraduate work, but it can be a good way to evaluate the learnings and not just the deliverables.”

Josh continued, “We have to think about the students we’re sending into the workforce. With AI, the expectation will be that they’re able to produce outputs at twice the rate. If they’re not familiar with these tools, we’re not setting them up for success. Supporting educators in the province to be at least on par with each other in how they employ AI and their approach to using it is essential to ensure future students get what they need from their post-secondary experience.”

Garbage In, Garbage Out

Prompting — the way in which we interact with most AI — is still in its infancy, but many users are finding and teaching ways to get better and more accurate responses from these tools. “There are three main types of prompts,” explained Trevor Simpson. “There’s the big one, where you write a paragraph about exactly what you want from the AI and hope you get the right answer. There’s a question-and-answer feedback method, or there’s the conversational prompt. All of those can work, but they generate different results from the various AI models. ChatGPT typically is the best to get scientific responses, but it frequently matches and mirrors your tone. If you’re in a sarcastic or relaxed tone, that’s the response you’ll get. If you ask in an academic way, it may over-talk and get lost in its own words. Microsoft Copilot needs context, or it may start hallucinating. And Google Gemini is still a bit inconsistent.”

“To show the different outputs and how prompts can provide necessary context, I tasked each AI to write a short poem about a sunset. They quickly delivered their poems, and each was decidedly different. The poem from Google Gemini was only four lines. But then I followed it up with the prompt, ‘Write a short poem and include features of the natural landscape of southwestern Manitoba.’ Then they started to include things like ‘golden wheat’ from ChatGPT or the ‘sunsets in the sky’ on Copilot. Again, Gemini was unimpressive. Finally, I asked them to write a short poem that included references to Southport, Manitoba, the Northern Prairie skink, and the Souris Swinging Bridge. And this time, Copilot won, saying, ‘Cross the Souris Swinging Bridge. Beautiful since 1904. It spans the river bold.’ I had to look it up and found out it was actually built in 1904. It was interesting that it took that leap in creativity.

“For the final part of the workshop, we explored interactive prompting, where you try to position the AI to think a certain way, such as ‘What is the best practice in writing a lesson plan for academic integrity? Who would you ask for that?’ The AI spits out best practices and said roles, including instructional designers, educational technologists, teachers, and librarians. I then created a prompt to say, ‘You are a team of those people. I need you to write a lesson plan based on its previous output, and the quality shot through the roof compared to a simple prompt of ‘give me a lesson plan.’ So, the power is really in the prompt, and the more context you can provide, the better outcomes you’ll receive. It’s important to understand that generative AI is still learning and will continue to be developed. It’s come a long way since its introduction to the global stage last year, but there’s still considerable room for improvement.”

The Human Factor

“Generative AI and program knowledge is a double-edged sword,” said Trevor. “You can use a lot of these tools to help maintain academic integrity and to watch for misconduct, but it will never replace the educator and the connections and conversations with their students. Generative AI can be used as a tool or a weapon, but it’s nothing more than either one of those, depending on how it’s laid out. We’ll always have the teacher-student relationship and trust to get through it.”

Keep Reading:

• Building Academic and Real-world Integrity in Manitoba
• Academic Integrity: Creating an Environment to Help Students Succeed
• Show Your Integrity – academic supports from the University of Manitoba