Recently, at a party with faculty from other institutions, I overheard someone mention that colleagues in her department weren’t going to share any resources with their students about AI because the faculty in question weren’t “on board” with it.

As CETL’s faculty fellow for AI and teaching this year, I won’t tell you to “get on board” with AI. What I will say, though, is that regardless of your personal feelings toward generative AI and its use in your courses, you can’t avoid talking to your students about it. The less you talk to your students, the more likely they are to use generative AI for unsanctioned purposes and to use it terribly.

“Age-Appropriate” AI Discussions

We might look at conversations parents have with their children about sex and drugs as a helpful parallel to the generative AI moment. In an article written for parents, the American Academy of Pediatrics (AAP) explains, “Some parents may not be comfortable talking with their children about sex and sexuality. However, if children aren’t getting the facts about sex and sexuality from their parents, they could be getting incorrect information from their friends or the media.” When it comes to talking to children about sex, the AAP recommends using honesty in response to questions of children from all ages, but different levels of details depending on the child’s age.

If you’re not talking to your students about generative AI, they will still learn about it and use it, but you’ll have lost the ability to help them become more responsible and mindful users. I recommend an interpretation of the “age-appropriate” approach. Your conversations with students will be dependent on the course number (whether introductory, capstone, or somewhere in between), the student’s academic experience, and the course’s goals. In a large 1000-level general education seminar, I can’t do as much of a deep-dive into how large language model (LLM) data sets are trained on copyright material without the copyright holder’s consent or compensation or the implicit bias of the data used to train OpenAI’s ChatGPT. In a small upper division seminar about technology and media, we can spend an entire class period on these issues. A course that relies on argumentation and evidence will necessarily have a different relationship to AI tools than a course in which students need to learn how to obtain objectively correct answers to problems or use precise formulae.

How to Have “The Talk”

In all courses, I recommend making time within the first or two weeks of class to have the basic talk:

Share your AI policies

To figure out what policy is right for your course, and consequently the right AI talk to have with your students, think about the kind of assignments you give and whether you want students to work with or avoid generative AI on them. Also keep in mind the course’s scope and size, which may affect your ability to engage in meta-learning conversations. If you’re not sure what your policy should be, CETL has an interactive tool that can help you figure it out, as well as a repository of sample policies from various instructors, courses, and institutions.

Remind students that generative AI is neither a search engine nor an encyclopedic repository of scholarly information. 

Generative AI results must always be verified externally. There are many variations of a flow chart to help students understand this floating around the internet, though the originating source comes from a UNESCO quick start guide on artificial intelligence in higher education (p. 6). These flow charts ask students if they need a result that is true and accurate and if they have the expertise to verify accuracy. If yes to the first but no to the second, a different digital tool is more appropriate.

Suggest or require students to take our new AI literacy course, developed by OU Libraries. 

The micro-course introduces students to how LLMs work and to some of the ethical issues, such as copyright and labor concerns, surrounding the creation and use of tools like ChatGPT. It can easily be completed as one homework assignment. Students who successfully complete the course will have a digital badge to share. 

Conclusion

As you “age up” your conversations with students about generative AI, you might talk to them about the ethics of data sets, the extreme toll tools like ChatGPT have on the environment, or how their ability to successfully prompt AI might shape their future employment. In its most sophisticated form, you might assign activities that ask them to learn how to work with various AI tools, critique the AI’s output, and determine which AI is most appropriate for which task.

For now, though, just remember that it’s important to talk to your students about AI, so they don’t learn about it on the streets. Have “the talk” as soon as you can.

References

American Academy of Pediatrics. “Talking to Your Young Child about Sex,” Pediatric Patient Education (August 30, 2022). DOI: 10.1542/peo_document142.

“The Authors Guild, John Grisham, Jodi Picoult, David Baldacci, George R.R. Martin, and Thirteen Other Authors File Class-Action Suit Against OpenAI.” AuthorsGuild.org, September 30, 2023.

“ChatGPT and Artificial Intelligence in Higher Education.” UNESCO.org, 2023. 

Hajikhani, Arash and Carolyn Cole. “A Critical Review of Large Language Models: Sensitivity, Bias, and the Path toward Specialized AI.” Quantitative Science Studies (2024). DOI: http://doi.org/10.1162/qss_a_00310.

Save and adapt a Google Doc version of this teaching tip.

About the Author

Bridget Kies is Associate Professor of Film Studies and Production at OU and CETL’s faculty fellow for AI and teaching. She is the co-author of the article “From Attributions to Algorithm: Teaching AI and Copyright in Media Studies” for the journal Teaching Media and the creator of the video lecture “How to Train Your Algorithm: Responsible AI in the Classroom.”

Others may share and adapt under Creative Commons License CC BY-NC.

View all CETL Weekly Teaching Tips. Follow these and more on Facebook and LinkedIn.