Memo to All Teaching Faculty - January 25, 2023 AI & ChatGPT Guidance for Teaching THE OFFICE OF THE DEAN OF THE COLLEGE THE OFFICE OF THE DEAN OF THE GRADUATE SCHOOL TO: All Teaching Faculty FROM: Jill Dolan, Dean of the College; Rod Priestley, Dean of the Graduate School RE: AI & ChatGPT Guidance for Teaching January 25, 2023 Dear Colleagues, We imagine you’ve been reading as much about generative Artificial Intelligence (AI) lately as we have, especially the new program, ChatGPT, which some observers suggest might have radical consequences for teaching and learning, especially at colleges and universities. Despite the copious handwringing in the media and elsewhere about this chat bot and its implications for higher education, we and our colleagues in the College and the Graduate School remain confident that our undergraduate liberal arts and graduate education programs will remain vital, vibrant, and useful in the years ahead. If anything, AI will make higher education and the nuanced and sophisticated ways of thinking it teaches even more essential. The ability of AI tools like ChatGPT to generate comprehensible text and code makes close reading and careful discernment even more important for our students and ourselves. We expect that faculty have questions about ChatGPT, and we write now to provide some guidance. We also want to acknowledge that this field is developing exceptionally quickly. We’ll be learning about AI’s power to construct essays and assignments now, next semester, and for many years to come. Princeton University doesn’t intend to ban ChatGPT or to levy a top-down edict about how each instructor should address the AI program in your classes. We outline here instead some ways of thinking about this new landscape. First, do remember that our Academic Regulations clarify that students are expected to properly acknowledge their sources, adhere to course collaboration policies (which set the standard for “permissible” collaboration), and not seek an unfair advantage over other students. The undergraduate Honor Code and all campus academic integrity rules are quite clear that students must produce original work. With this broader context in mind, do bring to bear some discernment as you contemplate your course needs and your pedagogy. You might, for example, decide that using AI/ChatGPT violates your collaboration policy, or you might decide it’s appropriate for students to use in certain cases. Other suggestions for those of you who intend to address the chat bot with students: Be Explicit/Avoid Misunderstandings Be explicit about your AI/ChatGPT policy in your syllabus, on Canvas, and as you introduce assignments during your first (and perhaps subsequent) class meetings. Make your AI/ChatGPT policy as clear as your academic integrity and collaboration policies and put them on your syllabus where students will be sure to see them. Sample statements that might be useful are linked here. Explain Your Pedagogical Rationale Explain to students how completing the assignments in your course will help them learn, and let them know why their own, original work on these assignments is important to their education. Develop grading standards based on your pedagogical values or learning goals, and share them with your students. You’re welcome to consult with the McGraw Center or the Princeton Writing Program as you think through your standards. Design Assignments with Care To minimize the risk of academic dishonesty, design assignments that require critical and creative thinking (rather than simply summarizing information). Ask students to demonstrate their process and reflect on their work. For example, they might annotate their solution to a novel problem or write a cover letter for their essay that introduces their ideas. Scaffold their work with multiple draft and revision deadlines that offer you opportunities to give them feedback. Ask them to present drafts of their work verbally in class or track their citations in early versions of their work. Teach Your Students What AI Can and Can’t Do ChatGPT generates patterns of text on a probabilistic basis (similar to “autofill” in your email client, though more powerful). For this reason, it should not be listed as a co-author on scholarly work; we see it as a compiler rather than a writer (this article from Nature addresses this issue persuasively). ChatGPT doesn’t have access to web material behind paywalls and has little knowledge of events after 2021. The program can’t (yet) cite references. Helping your students recognize what ChatGPT does and doesn’t do—as well as its potential benefits and risks as students construct their own original work—is valuable. Among the most promising teaching suggestions we’ve seen so far: After students have read and discussed assigned texts, invite them to work in small groups to ask ChatGPT questions about the readings and then assess and critique its answers. After students complete their own drafts of an assigned essay, invite them to request a draft of the assignment from ChatGPT. Facilitate a discussion in which students analyze how the drafts compare, for better and for worse. If they want to incorporate any aspect of the ChatGPT draft into their own revision, offer them guidance on appropriate citation and acknowledgement practices that will make their use of the tool transparent for readers. Again, ChatGPT cannot be listed as a co-author of scholarly work. As we know, ChatGPT won’t be the last AI program that challenges authentic teaching and learning. But this challenge is one we think Princeton faculty and instructors can manage with grace and aplomb. We hope you’ll share your teaching suggestions with us as you experiment with this emerging technology, and keep us posted about what you’re seeing and hearing in your classrooms. McGraw will facilitate a discussion about ChatGPT on Zoom sometime early this semester. Warm best to you all, CC: Liz Colagiuri, Deputy Dean of the College Lisa Schreyer, Deputy Dean of the Graduate School Kate Stanton, Associate Dean of the College, Director, McGraw Center Amanda Irwin Wilkins, Director, Princeton Writing Program Kathleen Deignan, Dean of Undergraduate Students Sample Syllabus Statements on AI/ChatGPT We encourage faculty to be explicit about their AI/ChatGPT policy in their syllabus, on Canvas, and during class. We offer two sample syllabus statements below: Intellectual honesty is vital to an academic community and for my fair evaluation of your work. All work submitted in this course must be your own, completed in accordance with the University’s academic regulations. You may not engage in unauthorized collaboration or make use of ChatGPT or other AI composition software. Students must obtain permission from me before using AI composition software (like ChatGPT) for any assignments in this course. Using these tools without my permission puts your academic integrity at risk. Further Reading Articles and essays about ChatGPT Bogost, Ian. “ChatGPT Is Dumber Than You Think.” The Atlantic, 7 Dec. 2022. Grobe, Christopher. “Why I’m Not Scared of ChatGPT.” Chronicle of Higher Education, 18 Jan. 2023. McMurtie, Beth. “AI and the Future of Undergraduate Writing.” Chronicle of Higher Education, 13 Dec. 2022. McMurtie, Beth. “Will ChatGPT Change the Way You Teach?” Chronicle of Higher Education, 5 Jan. 2023. Marche, Stephen. “The College Essay is Dead,” The Atlantic, 6 Dec. 2022. Marcus, Gary. “AI's Jurassic Park moment.” The Road to AI We Can Trust, 10 Dec. 2022. Prochaska, Eric. “Embrace the Bot: Designing Writing Assignments in the Face of AI.” Faculty Focus, 23 Jan. 2023. Roose, Keven. “The Brilliance and Weirdness of ChatGPT.” The New York Times, 5 Dec. 2022. Sharples, Mike. “New AI Tools.” LSE Impact Blog, 17 May 2022. Stokel-Walker, Chris. “ChatGPT Listed as Author on Research Papers.” Nature, 18 Jan. 2023. Warzel, Charlie. “Talking to AI Might be the Most Important Skill of this Century.” The Atlantic, 8 Feb. 2023.