Building AI policies into your syllabus

No matter your position on the use of so-called AI in the classroom, I recommend that you consider first your course goals and whether and how any given technology might undermine or support your students in the process of meeting those goals. In full disclosure, I take a “critical AI literacy” (Bali 2023) approach to the media-generative tools now commonly called “generative AI”: that is, we need to understand these tools in order to know when and when not to use them, and we need to consider them in the larger context out of which they emerge, including the harms associated with them. My approach below is informed by my “Blueprint for an AI Bill of Rights for Education”. Just as a reminder, Sean Kamperman and I put together some curated resources on the topic of critical AI literacy for our AI & Digital Literacy Summit in June, including a number of tools (NB: assume no particular judgment on our part on the inclusion or exclusion of any given tool). 

For your reference and to give you a sense of a range of possible individual classroom policy responses to generative AI (the cluster of tools including ChatGPT and other LLMs like Bing, Bard, and Claude), Lance Eaton has curated a collection of “Classroom Policies for AI Generative Tools.” (Note that many of these were crafted in early spring 2023 before the launch of other models and before the incorporation of generative AI options into Google Docs and Grammarly [known as GrammarlyGo, and currently only available with the paid model]).

Below is a short draft syllabus policy that you are welcome to adapt and/or adopt; it is not meant to be prescriptive. There may well be uses for AI tools in your course, and if you can imagine ways that students might productively use them, I recommend being very clear about where and why students might be allowed to use them and how you want students to cite their usage.  (Note: there is not yet an agreed-upon standard for citing large language models; the APA has suggested a model, but it has been criticized for treating LLMs as equivalent to authors—problematic for a number of reasons including that they synthesize the work of human authors without citation and that their outputs are not repeatable.)

I have also included the draft of my longer policy for the class in which I will be teaching about generative AI; this includes language about citing use (and distinguishes between researching and analyzing generative AI and using it to complete writing assignments). If any of it is useful for you, feel free to use it. I am also happy to discuss anyone else’s draft policies if you would find that useful.

Sample generative AI policy

Academic integrity is essential for a fair evaluation of your work and that of your classmates, as KU’s policies on academic misconduct make clear. All work submitted in this course must be your own.   

Assignments for this course are designed to help you develop your critical thinking, close reading, research, and writing skills. Submitting other people’s work as your own, repurposing work you have done for other courses without my prior approval and without substantial revision, or using so-called AI tools for your work in this course undermines those goals. In this course, you may not use ChatGPT or other generative AI software for your assignments, including plug-ins or native tools that automatically generate text within word-processing apps. Exceptions include the grammar check and spellcheck options on your word-processing software or Grammarly grammar check (but not GrammarlyGO’s generative text option); these approved tools make suggestions for revising work that you have already generated and may be used with caution and with the understanding that you are, as always, ultimately responsible for the work you generate. If you have questions about what constitutes academic misconduct (plagiarism or unauthorized use of tools or assistance), please ask me before submitting your work.

Draft, full academic integrity policy for my HNRS 190 on generative AI

This course is about so-called generative AI and aims to build literacy in concepts related to those tools, including understanding the ethical and social implications of this technology and what might be appropriate uses for it. It is likely that you will choose to research and analyze generative AI outputs as part of your final project; if you do so, as part of that assignment, you will disclose which tools you have used and explain how you have used them. You should keep a complete, accurate record of your prompts and the model's responses; the name and, where available, version numbers of the model (e.g., Chat GPT 4, Midjourney 5.2, NightCafe SDXL 0.9, Bing Chat Creative Mode); and the dates of your interactions (which will, among other things, help us reconstruct versions for those models that don't provide them, like Bing; but will also provide a more nuanced sense of models that are fine-tuned between version releases). Keeping detailed records is good practice for any course in which the use of generative AI is approved for any portion of an assignment.

Analyzing generative AI tools and their outputs as part of an assignment is not the same as using them to complete your assignments. At this stage in your educational career, I aim to help you develop your critical thinking and research skills, no matter what your chosen disciplines/field(s) of study might be. I am also interested in helping you develop your ability to communicate your thoughts and opinions. For these learning goals in this course, using (rather than analyzing) generative AI in your assignments will not help you gain the competencies you need, and indeed are likely to undermine them.  With this in mind, for writing assignments for this course, you are allowed to use built-in spellcheck, grammar check, or Grammarly (without the generative AI feature) after you have already brainstormed and drafted your work. You should not, however, use any separate, built-in, or plug-in tool or feature that generates text (for example, Wordtune, Google Docs "Help me write," GrammarlyGO generative text, ChatGPT, Claude, Bing, Bard) during the idea-generation or drafting stage of the writing process.  Ultimately, remember that you are responsible for the writing and work you share, no matter what course you are in or platform you use to communicate, so even if you are using approved tools, check your work before submitting it.

Unauthorized use of AI tools in generating work for this course and attempting to pass it off as your own work constitutes plagiarism; but plagiarism is a broader issue and is not dependent upon AI. Some specific examples of actions that constitute plagiarism include pasting together uncredited information or ideas from the internet or published sources; copying the wording of another source without quotation marks--even a few words--to indicate that the language is not your own (even if the source itself is cited); submitting an assignment written by someone else; submitting a paper written for another class without my approval and without substantial revision; and copying another student’s work (even with the student’s permission). In order to avoid unintentional plagiarism and to represent your work honestly, you will need to be meticulous about giving credit to any and all sources, whether directly quoted or paraphrased. 

Plagiarism hurts you as much as it does anyone. If you plagiarize work, you will not be receiving the practice and needed feedback to improve your own critical thinking, research, and communication skills. Plagiarism will generally result in a 0 for the assignment in which the plagiarism was found; I will notify you and make clear my concerns, and you are always welcome to discuss them with me.  All confirmed incidents of plagiarism will be penalized, reported, and kept on file in the English Department, the College of Liberal Arts and Sciences, and the University Provost’s Office, which means that the consequences of a second offense are higher. If you have questions, please ask me before submitting your assignment. Asking questions in advance of submission about plagiarism or academic misconduct will not be held against you in any way.