Academic Integrity & AI Course Policies

There is no central university policy on using AI tools. Yale College and each of the graduate and professional schools set their own standards, and most of these policies leave it to the individual course instructor to decide. The Yale College regulations discuss AI use in the context of plagiarism and unauthorized collaboration, and failure to follow a course’s AI policy is treated as an academic integrity violation. Students have been suspended and either failed courses or seen their grades lowered for misusing AI.

Annotated Readings of Example AI Policies

The Poorvu Center does not set or enforce policy

But we are in conversation with faculty from across the university. In our guidance for instructors, we advise them to develop detailed AI policies that are appropriate for their courses, and to discuss these policies with students. The examples below are meant to help support these discussions. In each, we take a section of a course AI policy and try to surface some of the underlying priorities or challenges students might face in applying the policy to their own work.

AI Policy Author Example AI Policy Annotated Reading of Policy
Mira Debs, Lecturer in Sociology and Executive Director of Education Studies, EDST 110: Foundations in Education Studies 
  • In this class, any written work that you submit at any stage of the writing process—thesis, outline, draft, bibliography, final submission, blog posts, and more—must be your own ideas and writing, and not copied or AI generated.  
  • Importantly, you may use AI as a research and editing tool. Any use of AI such as Chat GPT to support work on an assignment must be acknowledged in a citation that includes 1) the prompt you submitted to the bot, 2) the date of access, and 3) the URL of the program. 
     
 This policy draws a sharp line between using AI to generate writing (which is prohibited) and using it as an editing tool (which is permitted). Research supports this distinction, and in general you will find that if you introduce AI later in the process, you will retain more of your learning and authorship. In practice, the difference may not always be clear. If an LLM suggests synonyms or rewords unclear phrases—is that editing or generating? If the policy in your class distinguishes writing or generating text with AI from editing or getting feedback, we have two suggestions to help you keep track of the difference. (1) Be sure to cite the fact of your AI use, probably in a footnote where you describe what kind of help you received. (2) Keep separate and clean copies of the writing you submitted to AI and the writing you turned in for the assignment.
Charles Bailyn, A. Bartlett Giamatti Professor of Astronomy and Physics, ASTR 180: Introduction to Relativity and Black Holes 

Students may work together on problem sets, but the work submitted should be the student’s own. What that means is that when you are writing up what you hand in, you should be alone, and unprompted by someone else’s work (that is to say, not looking at anything other than your own notes).   

The use of ChatGPT or other generative AI follows the same rules: while AI may be used as a source of ideas, text generated by ChatGPT etc may not be submitted. We submit all questions to ChatGPT before the assignment is circulated, so we know what those answers look like. That said, for the problem sets we will not go to great lengths to uncover such violations, and we won’t worry about ambiguous situations. The purpose of problem sets is to learn the material, and to practice answering the kinds of questions that will appear on the tests. So while it might be possible to spend less time and get better grades on the problem sets themselves by submitting the work of others (human or artificial), in the long run this approach will result in the student spending more time and/or or doing less well on the course as a whole, and is thus a losing strategy.

 This policy emphasizes the learning loss that results from not struggling through the problem sets on your own. It also explicitly tells you to write answers in your own words, even if you collaborate with AI or other students in thinking about the problems. This directive is well supported by educational research showing that trying to explain things in your own words teaches much more than reviewing other people’s explanations—whether a classmates’, your lecture notes, the textbook, or an answer generated by an LLM.
Michael Farina, Lector I, Italian Department, ITAL 110 Elementary Italian I  Use of Generative AI is both permitted and encouraged. You may also use our (beta) TutorBot to help you learn Italian (www.yaleoer.com). You may not, however, use AI systems like GPTs to complete your work for you with little to no active participation on your end. In general terms, you may use AI to help with your work. You may not use it to do your work. Use of AI should be an active, or interactive, process that helps you learn. It should not replace learning with passive production of assignments and other work. If it doesn’t feel right, it probably isn’t. When in doubt of AI-use on assignments, a best practice would be to ask your instructor if it is acceptable use. Any inappropriate use of AI to pass off generated content or work that is not your own will be treated as academic misconduct. Like the STEM example, this policy distinguishes using AI to learn and practice (which is permitted) from using it to generate work you will submit (which is prohibited). What’s especially striking here is the advice to check your gut sense of whether a given AI use is appropriate, guided by a reminder that learning needs active engagement. In the long run, working with AI in a variety of ways and reflecting on how that feels is a solid goal for developing your own sense of when it helps and when it hinders learning.

Frequently Asked Questions

At a minimum, follow MLA (or another style as required) guidelines. If a course or assignment requires you to cite AI and you are unclear how best to do so, please ask your instructor.

Assume no and err on the side of caution. Asking your instructor is an opportunity to better understand the assignment and the knowledge, skills, and abilities it is intended to promote.  

Assume yes if you are not using a Yale approved tool. Faculty cannot submit your work or personal information to a non-approved tool; you cannot submit course materials to non-approved tools.