Integrating AI in Education

Educators who attempt to ban students from using AI tools in all aspects of their coursework are engaging in a losing battle. Moreover, this approach is a disservice to their students, who will likely be expected to know how to use AI tools responsibly and effectively in their future workplaces. Dr. Sarah Elaine Eaton of the University of Calgary supports this viewpoint, stating that AI tools “can be used ethically for teaching, learning and assessment,” and that student use doesn’t necessarily constitute an academic integrity breach.^[1]

AI is here to stay and will be integrated into most aspects of education. Accordingly, institutions, departments, and instructors need to set policy and clarify learning outcomes, and provide students with clear expectations on AI use for every assignment.

Setting Institutional Policy

An October 2023 report from the Canadian Digital Learning Research Association recommends that institutional policies should be developed that promote the “effective, creative, equitable, and responsible use/nonuse” of AI tools and promote staff learning and experimentation around AI tools.^[2] See the table below for a selection of AI policy/guideline highlights from Canadian polytechnics. Clicking any link in the table will download a PDF copy of the institution’s policy statement.

Educators and students should check with their institutions for emerging policy and guidance on appropriate use of AI tools. For a roundup of AI policies at educational institutions, visit Higher Education Strategy Associates’ Observatory on AI Policies in Canadian Post-Secondary Education.

Engaging Students on the Topic of AI

Discussion around AI tools should begin early in a course, ideally in the first week when introducing the course outline. This raises student awareness that AI tools are a part of the educational landscape, and positions the instructor as one who is aware of AI’s capabilities and is willing to talk about appropriate usage.

Creating a Course AI Policy

Course AI policies set clear expectations about acceptable and unacceptable uses of AI tools. Policies should clearly state how AI tools can be ethically used for coursework as well as procedures for documenting AI use on assignments.  Depending on institutional and departmental protocols, there may be a set policy that instructors need to share with their class, or instructors may be able to set their own or co-create a policy with their students. In any case, instructors must ensure that their course’s AI policy aligns with their institution’s policy on AI use, and are also responsible for following institutional policies regarding consistency across different sections of a course and courses within a program.

Several RRC Polytech instructors have reported success in engaging classes in establishing a course AI policy. Two RRC Polytech communication instructors conducted activities in which the classes were divided into groups and asked to draft lists of acceptable and unacceptable uses for AI tools on upcoming assignments. Subsequently, each group shared their lists, and the class collaboratively developed an AI policy for their course. In some cases students set stricter policies than the instructor would have chosen.^[3] Indeed, many students want to engage with AI tools but don’t want to miss out on developing critical thinking and other key skills.^[4]

Both instructors found that co-creating an AI policy led to an increase in dialogue around AI tools throughout the course and a decrease in AI-based plagiarism compared to the previous term.^[5] Research shows that engaging students in collaborative decision making around expectations leads to better uptake of a class policy, a more positive classroom culture, and increased engagement.^[6]

See the following text box for an example of a collaboratively drafted AI policy.

Setting Clear Expectations on Assignments

Students need to be presented with clear expectations about how AI tools can be used on each assignment. For courses with an established AI policy, students could be asked to revisit the policy once a new assignment is presented, and report back about how AI tools can and can’t be used. Another approach is to include a ‘stoplight header’ on each assignment that indicates how AI tools can be used, with red representing no AI usage on this assignment, green indicating that AI tools are allowed, and yellow meaning to ask the instructor before using AI tools.^[7] It’s also possible that an assignment could have red, yellow, and green guidelines for particular tasks within the same assignment.

Many course and institutional policies around student use of AI tools require students to disclose the role that AI tools played in creating an assignment. This might involve a simple declaration in the assignment or footnotes that explains how AI was used. Theresa Senft of Macquarie University has required students to submit full documentation of AI usage on assignments, including the prompts used, information gained, and explanations of how that information was used. This task was designed to highlight the importance of documentation in AI usage, considering the numerous examples of AI-generated hallucinations and errors resulting in employment termination.^[8] This type of exercise has several other benefits—it gives students an opportunity for feedback on their prompting and AI literacy, promotes critical thinking about how AI outputs are adapted and used, and models responsible use of AI tools.

The image below shows the beginning of Theresa Senft’s AI tools documentation template. Click the image to view the full template.

Clarifying Learning Outcomes

The rapid rise of powerful AI tools has reminded instructors of the need to critically consider how assignments connect to learning outcomes and real-world tasks. It also serves as a reminder to engage in dialogue with students about the learning outcomes for each assignment, why they are important, and how they will prepare students for the workplace. This can be useful alongside each new module or theme and as each assignment is introduced. Carly Schnitzler of Johns Hopkins University suggests exploring the importance of course learning outcomes while crafting a course AI policy, with a provocative discussion question like “What’s the point of being in a writing class, if language models can generate passable text and score well on exams?”^[9]

With the power of today’s AI tools, learning outcomes may need to be adjusted to ensure that they reflect skills that students will need in the workforce and that AI tools are not able to reasonably demonstrate. Oregon State University suggests an updated version of Bloom’s Taxonomy, which focuses on “distinctive human skills” that go beyond the capabilities of AI tools.^[10]

The following image shows Oregon State University’s suggestions for updating Bloom’s Taxonomy. Click the image for more information, or download a PDF copy of Bloom’s Taxonomy Revisited.