FAQs: AI & Teaching Challenges

Articulate the learning goals in your course that AI use might interfere with. Be explicit with your students about what they may lose when they undervalue or circumvent certain parts of the learning process by relying on AI. Connect your course objectives to specific skills that you want your students to learn independently, and explain why productive struggle is essential to growth and learning. 

Demonstrate AI’s weakness at a discipline-specific task. Here are some examples of “AI gone wrong” that you can point to for students to illustrate the negative impacts the tool can have. Or, seek out other examples more specific to your field.:

• Air Canada must honour refund policy invented by chatbot, court rules: A chatbot hallucinated a bereavement fare policy that didn’t exist, and a court ruled the advice was binding. 
• Generating Medical Errors: GenAI and Erroneous Medical References: AI models are performing poorly, in one instance 30% of individual statements were unsupported by the provided citations, and nearly half of its responses contained at least one unsupported claim.
• Fake viral images of an explosion at the Pentagon were probably created by AI: There have been numerous instances of AI-generated images going viral and being mistaken for real news, such as th​​e above story about one of an explosion at the Pentagon
• Amazon scraps secret AI recruiting tool that showed bias against women: AI tools can perpetuate historical biases, as in this case, where a recruiting tool penalized resumes that include the word “women’s” and downgraded graduates of all-women’s colleges.   

Name for students the many reasons to think critically about their AI use, which includes not only concerns about idea-generation and cognitive off-loading but also ethical, environmental, and moral stances. This compendium of “Teachable Readings” can help frame issues for students. Instead of thinking purely about use or resistance, help students consider ways to limit the resources their AI use consumes—for example, producing images or videos is more intensive than text-generation. 

Both educators and students raise valid concerns about the broader impacts of AI, and exploring these concerns with your students as they relate to your course and discipline can be a fantastic learning opportunity. These concerns generally fall into three categories: 

• Ethical & Intellectual Property: GenAI models are trained on massive datasets that often include copyrighted work, proprietary research, and creative intellectual property without the creators’ explicit consent, compensation or credit.
• Environmental impact: Training and running GenAI platforms requires massive data centers that consume excessive amounts of electricity and water for cooling systems and lead to pollution. (Learn about the pollution of AI and data centers and AI's threat to natural resources).
• Moral & Social Bias: Because AI models learn from existing human data, they can amplify and perpetuate historical biases, stereotypes, and misinformation. Additionally, there are moral concerns regarding the potential displacement of human labor, the erosion of critical thinking skills, and the “homogenization” of ideas. 

If you are requiring the use of AI for an aspect of your course and a student objects to using AI on ethical, environmental, or moral grounds, it is highly recommended to respect their stance and provide an alternative. Forcing usage can alienate students who are trying to align their academic work with their deeply held values. To navigate the situation, you might:

1. Validate their stance and remain curious: Acknowledge and validate their concerns, demonstrating you value their critical thinking and ethical engagement. “I appreciate your thoughtfulness. These are certainly complex issues the academic community is actively grappling with, and I’m interested to learn more about your concerns.”
2. Focus on the learning objective: Identify the core skill the AI-integrated assignment was designed to teach. Explain why the assignment includes the use of AI and what you hope for students to learn from using AI.
3. Offer an equitable alternative: Provide a path that allows the student to meet the same learning objectives without AI. For instance, if the assignment asked them to use AI to critique a draft, encourage them to utilize the Writing Center or do a self-critique using a rubric. Or, if the purpose of the assignment was to better understand how AI works, can you instead go through a demonstration together rather than asking each student to use AI? In considering alternative paths, you may realize that your requirement for each individual to use AI could instead be a recommendation or a group activity that reaches the same goal. 
4. Turn the resistance into a reflection and learning opportunity: If appropriate for your course learning goals, offer the student an option to write a brief reflection or analysis on the reasons they are choosing not to use the tool, mapping their objections to course themes. Consider holding a discussion with the class and engaging in exploration of the students’ views. 

AI detection tools are found to be unreliable and present significant equity concerns. Unlike traditional plagiarism checkers that match text to existing databases, AI detectors use models to guess if a text "looks" like machine output. False positives are common, where student work is incorrectly flagged as AI-generated. There is research to show that AI detection tools can be biased against non-native English speakers, likely because they often use more structured or formulaic sentence patterns to ensure clarity. Overall, relying on detection tools can create an adversarial environment that undermines student-faculty trust and could potentially lead to wrong accusations based on flawed data. 

Still, we recognize that academic integrity is a significant concern right now and faculty are looking for solutions to uphold high standards for authentic student work. Instead of trying to catch students, we recommend strategies that make it less feasible and less tempting to submit AI-generated work. At this point in the semester, it is likely challenging to significantly adjust your assignments, but some options to use as students turn in assignments are:

• Ask students to write in Google Docs and share the version history.
• Have an oral check-in where students explain their process and thinking.
• Ask students to hand write a summary of their main argument in class after submitting the digital version.
• Do a citation audit. If students include fake/falsified sources, that is the academic integrity violation you report, rather than AI use.

Effectively, no. A course-level policy that prohibits AI entirely is a policy that is nearly impossible to enforce and can introduce inequity, anxiety, fear, and mistrust. At this point in AI’s development, students can't avoid interacting with AI if they are on the internet. AI is built into email, web searches, and platforms that most people use on a daily basis, sometimes without even realizing it. Also, students are using AI to study, organize their time, review class notes, and in other ways that support their learning separate from completing assignments. However, you may decide that the expectation in your course is that students should not use AI to complete assignments or assessments.

If you have a strong argument for why students need to complete assignments and assessments for your course without the use of AI in any way, you should be clear about (1) what is prohibited (2) why and (3) what the consequences are. When explaining to students, it helps to pivot from “policing” to “mentoring.”

Here is an example of a syllabus statement that does not prohibit the “use of AI” but prohibits submission of work that was generated by AI:

By submitting work for evaluation in this course, you represent it as your own intellectual product. You may not submit for evaluation any content (e.g., ideas, text, code, images) that was generated, in whole or in part, by Generative Artificial Intelligence tools (including, but not limited to, ChatGPT and other large language models). Relying on AI prevents you from developing the skills that this course is designed to teach, and my goal is to help you find your voice and sharpen your critical thinking. Submission of AI-generated content will be treated as academic dishonesty, which may result in a failing grade for the assignment and/or a referral to Conflict Education and Student Accountability (CESA). 

Your expectations with students should align with your expectations for yourself. Being transparent about your own AI use can be a powerful learning tool. For example, you might explain how you balance the need to learn about AI with the knowledge that AI use consumes resources. 

To highlight to students that you are not asking them to use AI differently than you yourself would use it, you can use your policy to articulate your own use as well as your guidelines for students. Here’s one example of a policy that does this: 

• As your instructor, I will: 
  • Trust you regarding your use or non-use of generative AI tools. If you do not submit a reflective statement, I will trust that you did not use any generative AI tools.
  • Ask you if I have a question about your work. I get to know students through their class contributions and writing, and if you turn in something that doesn’t sound like you, I will ask that we talk about it together. 
• As your instructor, I will not:
  • Submit your work to any tool that claims to detect AI-generated text. These tools do not work: as one example, they flag the work of non-native English speakers as “AI-generated” at a higher rate than the work of native speakers.
  • Use a generative AI tool to grade your work or provide written feedback on it.
  • Use a generative AI tool to generate course materials.

To make sure the interaction is a learning experience that maintains trust between you and the student, approach the student with curiosity rather than accusation. There might be a misunderstanding or personal struggle you aren’t aware of, and approaching students with the intent to gather information will better position you to uncover and help address the root causes of the student’s actions. Reach out privately and keep the tone low-stakes. For instance “I was looking over your last submission and noticed a few things I’d love to chat about with you.” For a script that can guide this process, see “What to do when you suspect plagiarism.”

During the meeting, focus on the process rather than the penalty. Ask the student to walk you through how they approached the assignment or how they gathered their sources. Often, students who take shortcuts do so out of panic or a lack of understanding of the rules. By letting them do the talking first, you create a space for transparency and learning from their mistakes. If it turns out they did violate a policy, you can then pivot to a discussion about why the policy exists and how they can move forward productively. If you conclude that there are clear academic integrity violations, such as fabricated sources or unquestionable plagiarism, you can then talk through the charge of academic dishonesty you plan to report to CESA, which may be just a warning or a sanction like failure of the assignment.