About ChatGPT

Image by frimufilms on Freepik

Rifat Salam, Social Sciences, Human Services and Criminal Justice Department

But even when the essays are a good synthesis of other essays, written by humans, they are not human. Frankly they, they creep me out precisely because they are so competent and yet so very empty…The essay does not invoke curiosity, or any other emotion. There is a voice but it is mechanical…That’s because real voice is more than grammatical patternmaking.

–Tressie McMillan Cotton, NYT, 12/20/22

While generative artificial intelligence (GAI) technology has been around for some time, ChatGPT’s introduction in November 2022 and its rise to prominence is what brought it to most of our attention.  I’ve observed the use of GAI by students for a number of years, particularly in discussion board posts but the sophistication of ChatGPT and the quality of its products far surpasses its predecessors and professors who assign any kind of writing have to pay attention. As one of the co-coordinators of Writing Across the Curriculum at BMCC, I knew that this was an issue that faculty who teach Writing Intensive sections would be especially concerned about so we proposed a WAC/WI Refresher workshop in spring semester 2023.  In this article, I’ll discuss my first encounters with GAI and what we discussed in the workshop.

At the end of fall semester 2022, coinciding with the introduction of ChatGPT, I noticed a small number of submissions which seemed oddly generic, didn’t address assignment prompts nor did they meet the minimum requirements in the grading rubric.  Very much like the quote from Tressie McMillan Cotton, the submissions felt oddly hollow, only tangentially addressing the assignment guidelines and fitting a template rather than reflecting ideas generated by authentic learning in the class. Anyone could have written it, and you did not need to have taken Professor Salam’s Understanding Technological Society class at BMCC to know this. Indeed, if you did take the class, do the readings and complete the learning activities, you would not have written these essays. They did not meet the requirements on the rubric and received failing grades, with copious feedback explaining why.  I did not accuse the students of plagiarism or using AI—I did not need to. The majority of students submitted their own work, imperfect but full of evidence of learning. It will not surprise you to learn that those alleged AI submissions came from students who rarely attended class and when they did, their presence was cameras off with no participation.

So what are the implications for teaching and learning in the GAI era? In WAC, we have had multiple workshops over the years about creating “plagiarism-proof” assignments, and the same basic principles apply for addressing GAI. I have come to think of it as creating “human-oriented” assignments or more specifically, assignments that reflect the learning of the actual human students who take my classes.  Here are some highlights from the workshop:

  • Create a GAI policy to use in your course and include it in the syllabus. The college’s plagiarism policy does not strictly reflect the nature of GAI so we should not rely on it as a way to deter its use.
  • Get curious about how your current assignments fare with ChatGPT. Run assignments and draft assignments through the program to see what you get and make changes accordingly.
  • Rethink “traditional” writing assignments like the five-paragraph essay and research papers that can lead to “data dump”—GAI are particularly good at these.
  • Design prompts that are specific to your course and your class and style of teaching. Provide excerpts, tasks, source material, images, etc. that give space and structure for demonstrating learning.
  • Use scaffolding toward a longer assignment, with writing done in class if possible. I do this on Zoom using tools like Google Docs, Jamboard or Padlet.
  • Make your assignments as “original” as possible, e.g. data collection, field observations or writing about niche sources not widely available online.
  • Have a strategy ready for addressing AI-generated submissions. I allow a resubmission but only after a discussion.
  • Find ways to use ChatGPT as part of an assignment or class activity.
  • Emphasize writing-to-learn and process over product—these WAC principles provide a strong foundation and rationale for designing and assessing assignments that encourage and require students to do their own writing.

Here is an excerpt from my GAI/ChatGPT policy that communicates that last point to students:

The writing we do in this class is designed to help you learn, to develop your thinking—it’s not about creating perfect paragraphs with no grammar errors though the draft and revision process can help with that—instead, the writing is about generating and refining our ideas. AI-generated writing is a tool that can have legitimate uses in many contexts but not in a WI or any college course, where writing is meant to be a tool for communicating your ideas and what you have learned.

In the class I teach, SOC 111: Understanding Technological Society, we discuss the impact of technological innovation on individuals and social institutions, and we address this not as an issue of cheating as much as a way that the technology becomes dysfunctional to the learning process. I asked ChatGPT to write an essay about why students shouldn’t use AI to write their papers. Here is the most important point it generated: “When students rely on AI to generate their content, they miss out on the valuable opportunity to engage with the material, form independent opinions, and develop critical thinking skills. In the long run, this can hinder their ability to think critically and problem-solve, which are essential skills for success in both academics and the professional world.” (ChatGPT 2023) The biggest thing I try to communicate to my students is why it is worth the time to do the writing and why I would rather read their “imperfect” words over cold, empty grammatical patternmaking any day.

References

Additional Resources:

Leave a Reply

More From This Issue