I am no stranger to dealing with AI invading every single aspect of my life; as an artist and writer, I have to contend with companies sweeping up my data and the constant discourse about whether I “should” be allowed to make money with my writing and illustration work. Especially as we are now in a world where people can generate passable images with some text prompts. Being able to create and having the audacity to charge for it paints me by some AI-enthusiasts as a gate keeper.
In writing, specifically, I have to tiptoe through a minefield of integrated AI “helpers” in researching information or the writing and editing process. These tools are “helpfully” integrated into every program I use. They often take up huge amounts of UI space, cutting into my flow state. Sometimes, they integrate themselves directly into my process without my asking.
As a podcaster I’ve spent hours researching and writing about generative AI and the weird things it does. I have yet to find any AI integration on a daily basis that truly makes my life easier. Nor have I seen generated content in an overall quality that is better than anything I could potentially do. It hasn’t really saved me time, either.
But AI in the classroom? That is by far the most intrusive, unhelpful aspect of all of this AI-bubble. It has been an active detriment to my practice. It is so bad I wonder if they is anything that can done?
My Background with AI
I write all this not as someone experiencing the sour grapes of generative AI affecting my livelihood, but as someone who has had a role to play in the development of generative AI. During times of desperation I found myself doing A/B testing and contributing to a couple of large AI companies as a freelancer. I am not sure if I am still under an NDA, but they rhyme with “Shmalfabet” and “Medduh.” It was enlightening to get a peek behind the curtain. I saw how data was processed, and the conditioning of the AI models regarding queries and safety issues.
And of course I’ve dabbled. Who hasn’t? I’ve prompted to see what can be generated. Yet, I never found results that were particularly compelling or useful to me. It just seemed faster to do things for myself and to my specifications. AI generators could never seem to quite reach what I asked for. Additionally, my output is more unique than what I was seeing generated. AI just isn’t for me, and I know that.
It’s for people who want to not have to deal with people like me. The promise of generative AI is to help the rich to eliminate the pesky human element of production. As long as the generative AI production pipeline can churn out material that is “good enough” for the population without resulting in profit loss, generative AI is deemed successful.
Unfortunately for me, however, the “good enough” has proven to be “good enough” for higher education as a whole. It is getting hard to combat the rising tide of AI generation in the classroom.
The AI Dam Breaks
This semester, specifically the past month of October, was the breaking point for me as an instructor.
I am someone who just wants to teaching students how to write; I teach them standards expected of upper-division college and have them practice those standards in the work, but I implement these things in intellectually interesting ways. Punk aesthetics, media, monsters… the intent is to provide students with a foundation in the academic standards of writing, but once mastered, they are free to indulge and experiment, all while writing about concepts that are outside of the English composition canon.
While I think I have been successful overall, the AI generation I have increasingly seen has just reduced the overall effectiveness of my philosophy in the classroom. Rather than mastering the essentials so that they can experiment, I am now finding myself dealing with students using AI who are not mastering the essentials and falling behind because the AI generative tools are “good enough” and not understanding why I point out their papers are flat, generalized, and boring.
My AI Experience In the Classroom
I have four classes I am currently teaching. I expect an AI paper or two per class. But in the most recent major paper assignments I had 20% of all submissions flagged as AI. And yes, I am aware false positives exist. I know enough to not blindly see these things as guaranteed AI-generated papers. But based on the internal metrics, analysis of writing style, and experience I have in processing thousands of AI responses, it was still 20% of submission that needed to be investigated. That takes time – a lot of time. Lord forbid you have a student who uses AI generators and wants to fight you on it, because how do you prove it?
You can’t. Not really.
You can point out that the writing has all the hallmarks of the LLM style, that sources are hallucinated, and that formatting is incorrect. But none of that constitutes “proof” today and students are more than happy to claim they are just bad writers.
I would find that mortifying, frankly.
So What Do I Do About AI?
I’ll let you know what to do about AI in the English classroom when I figure something out.
I kid, of course.
The reality is, until these generative AI modules feed on enough of their own output that they collapse, there is nothing that can be done, really. Outside of going back to handwritten assignments in general, I guess, but I doubt that would be worth the effort. Hell, a dedicated enough student could potentially just hand copy something prompted. Though you may wonder why go through that effort when they could just do the work themselves, do better, and learn something.
I also can still fall back on just grading for writing effectiveness. Flat, boring, and no personality? Then I can grade for that. Hallucinated sources? Bounce it back to the student to find the exact source and provide it to me. If they’re unable, they’ve clearly failed the research element of a writing project. Formatting issues? Failure to leverage the template and resources? That is a flat deduction because I have the tools there for you. Sources make no sense or seem not to exist? Please show me where it was found, as all sources must be properly documented. You lost it? Well, that sucks.
Moving Forward
In my own case, I am thinking a lot about moving critical thinking practices away from strictly writing. I have a few ideas to reduce the weight of traditional composition and lean more into projects that encompass the classroom concepts that lean into performative and aesthetic traits; things like presentations and producing zines that articulate argument and research as opposed to standard MLA-formatted papers. My thinking is that putting the practice of writing techniques into a different sort of craft can shake things up a bit.
The classroom I am increasingly leaning toward, even as I prep for next semester, is one of lower stakes, frequent check-ins. I am taking larger paper concepts, breaking them down into forms of writing that are more personality-driven and focus more on personal interpretation rather than research. Personalization, the quirks of the individual, and realistic “feelings” are things AI has yet to quite nail, and the idea is to shift that direction.
Of course, I still have papers – but the intent is smaller, lower stakes ones that are more specific and requiring material that may not be readily available to AI generators. The papers require more personal views and leveraging one’s sense of aesthetics, unique to them, not something generalized by an algorithm. I can only give them the tools and instruction, whether they are content to churn out garbage is up to them. It’s their money. But I feel I am doing my part in any case.
Of course there are ways around these things for an enterprising student who just doesn’t want to do the work. But they still will be churning out shit because they’re not learning anything. I can at least grade for that.
Thank you for reading, if you are an educator trying to figure out how to deal with this AI issue like I am, I would love to hear your thoughts! While I don’t have much regarding practical solutions, here, I hope my perspective might prove interesting.
