What's the Reality of Artificial Intelligence in the Classroom?

In research, it has been a reality for some time. Think Google, or other better curated digital resources that our colleagues in the library can help us think about and put to work. Colleagues in world languages have also been addressing for some time the use of Google Translate and other applications for translating language. I am told the same is true for coding in computer science.

Writing assisted by A.I. is also not new. But recent developments and improvements in a particular use of A.I. machine learning for natural language processing and text generation has received lots of attention in recent weeks. The tool is a chatbot called ChatGPT, based on software known as GPT-3 developed by OpenAI. It’s a tool that can enable and assist in research and brainstorming, organizing and outlining, editing, and even delivery of content. By many accounts, though not all, this is a tipping point. It can generate a text–an entire essay or article–that reads as though a human wrote it.

I am very interested in rhetorical tools–analog and digital–that we can use to generate and compose thinking. Aristotle didn't have a computer, but he did have technology. (In ancient Greek: techne, meaning art or know-how; it is also the root of our “technology” and “technique.”) In fact, when the recent AI update was released, I was in the middle of research that explores the ancient rhetorical practice and techniques of artificial memory and seeks to bring it into conversation with contemporary cognitive psychology and learning sciences for use in the classroom today. There are correlations. And also, I hypothesize, significant differences. The difference, I believe, has to do with the difference between artificial intelligence/memory/composition and automatic intelligence/memory/composition. That’s scholarship in progress.

I am intrigued and open-minded and also, in the scholarly sense, critical about the ways we (students, teachers, scholars) can use new digital tools to update the effectiveness of our heuristics and other rhetorical tools for the work we compose. I am concerned by the fact that what is being celebrated at least in part, and what is clearly being sold to students in some commercial quarters, are the wonders of writing that automatically and completely done for us, not just assisted by artificial intelligence. No learning or work required. There is a difference, then, between artificial and automatic intelligence that I want to explore in my scholarship and consider further in my teaching. It is worth noting that OpenAI does not expect the use of its generative software to be automatic; they expect users to attribute the use of the resource and emphasize that the writing was not automatic and not immediate, but reviewed, revised, regenerated by the human writer after using the mediation of the software. See its Publication/Sharing policy. Is this acknowledgement of the mediacy and non-automaticity of the intelligence generated through ChatGPT shared by other entities who use this or other varieties of AI? How much transparency can we expect? A glance at EssaySoft and how it markets its services for easy essay creation suggests the value of automaticity for otherwise busy students. That tech (I assume) is not be GPT-3. But do students understand the difference? What is the difference? How do we value the difference? These are the kind of questions we should be asking as educators and scholars. I have been dismayed to find these kinds of questions, or frankly any questions, lacking in more than a few iterations of frenzied exuberance that has emerged in print and social media in recent days.

I plan to bring these recent developments and issues with A.I. into discussion in my classes this spring, classes which foreground the work and process and "rhetorical knowledge" (one of our key Writing Program learning goals) that go into the composition of ideas and arguments. I want to provide some context for the approach I take with it, with regard to policies I identify on my syllabus and a practice I use when students turn in their essays. Both are copied below.

The consensus I’ve heard from our peers and colleagues across the country the past couple weeks on email lists and Zooms: AI-generated writing in higher ed is already here, like it or not. We can ignore it, fight it, celebrate it. Or educate for and perhaps around it when necessary. While doing so, we can also interrogate and think through it, as scholars. I am working on developing an initial discussion and workshop around this real issue–collaborating with colleagues in Computer Science, World Languages, LAT, and other places on campus. Stay tuned, and please be in touch with questions, suggestions, examples, and ways you would be willing to contribute in a future discussion. I’d also love to hear from anyone interested in potential collaborations on scholarship regarding any of these issues.

My Syllabus Statement:

Washington College has the following policy regarding academic integrity and plagiarism: Plagiarism is defined by the Honor Code as “willfully presenting the language, ideas, or thoughts of another person as one’s original work.” Turning in someone else’s work as your own is plagiarism. Relying on other texts, resources, and intelligence (both human and artificial) to generate our work is basic to what we do as scholars and writers. Acknowledging those resources is also basic to what we do. For that reason, quoting or paraphrasing or otherwise using the words or ideas of other people and resources (such as Wikipedia or ChatGPT) without properly acknowledging your source is also a problem. If you ever have any question at all about whether you are using a source correctly, ask me about it to learn more. (We will be talking and learning more about the rhetorical uses and potential abuses of artificial intelligence tools like ChatGPT). Submitting a paper for this class that contains all or part of a paper that you submitted in another class, without the permission of both professors involved, is also a violation of the Honor Code. A student found guilty of plagiarism may fail the assignment or the course, and may be referred to the Honor Board for further adjudication. Whenever you hand in a paper for this course, you will include as part of a preface an acknowledgement of the resources you have relied upon and a statement that your work has been completed in accordance with the Honor Code.

Essay Format Prompt:

The writing projects should be approximately 5-6 double-spaced pages/1500 words (not including Preface) in 12-point font, standard margins. [“Approximately” means that a project much shorter 5 pages or much longer than 6 is likely in need of further revision]. Each project will be submitted to Canvas. The final version uploaded to Canvas must include this preface on the document at the top, clearly separated from the main body of the essay. The preface includes:

1. Abstract: Provide the final version (3-5 sentences) of your argument’s GPR (given/problem/response).
2. Reflection: Identify at least one element of your writing (from the rubric and/or your to-do list) that you have focused on and believe is working in this project and one element that will remain on your to-do-list for further attention and feedback.
3. Acknowledgements: Recognize support received or relied upon such as: feedback from a classmate or friend, a Writing Center tutor, a professor, a librarian, a pet or family member; use of resources, including composing/editing software, not included in your Works Cited. And your affirmation of the Washington College Honor Code.

All citations (direct or indirect) should use MLA format or another that you are familiar with already–which includes a listing of the works you have cited or quoted from. For guidance on proper citation format [in-text citation; works cited at end] consult the Purdue OWL.

Here is a curated resource that I will be consulting as a guide to further reading and thinking:

Mills, Anna (Curator). AI Text Generators and Teaching Writing: Starting Points For Inquiry. (2022).

[Acknowledgements: I did not use ChatGPT to generate or edit this post; I used Google to retrieve the links to OpenAI and EssaySoft. I am grateful for the discussion and resources generated by a POD Network workshop on this topic. I am also grateful for the patience of my family while they waited for me to complete this post; it took some time.]