By David Gramling (he/they), professor and head of the Department of Central, Eastern, and Northern European Studies, and Matthew Chan, third-year undergraduate student in Arts.
It’s not an uncommon experience to feel anxiety about language(s) at university, especially in matters of technology and marking.
You might fear that the language you’ve written for an essay won’t live up to what your professor or Teaching Assistant believes is good enough. You might fear that, when you learn a new language like Danish, Catalan or Punjabi, your performance in it as a learner might not advance quickly enough to make you feel successful, sound good to your peers, or get a good grade. You might fear that you can’t control the way other people hear your voice, its particular sounds and shapes, or what some might call an “accent”. All of these fears may be well-founded based on your experience, and they frequently make people more hesitant when expressing themselves, building relationships, and trying out new things with language.
Since December 2022, there’s been a new beast on the block when it comes to linguaphobia, or anxiety about language: ChaptGPT. It has led to a bit of a frenzy among educators and students to snoop out which assignments have actually been written by real students in the class, and which ones are NARPy (written by Not A Real Person). Matthew Chan, a third-year student in UBC Arts, had a pretty jarring experience around this set of issues. We got to know each other when his written work was flagged as evidencing AI-generated features in a class in my department last year. Where other students who ended up in my office for this reason ultimately admitted they had used ChatGPT in a pinch to complete a late-night assignment, Matthew had not done so. He was confused and taken aback by the allegation.
This is part of an ongoing GenAI op-ed series, Arts Perspectives on AI (Artificial Intelligence), that features student and faculty voices from the UBC Faculty of Arts community.
So Matthew and I decided to do an experiment together to figure out what was going on. We both wrote essays by hand with pen and paper for 30 minutes at my office table, on the 9th floor of Buchanan Tower, on an identical topic chosen for us shortly before. My CENES Nordic Indigenous Studies colleague, Professor Dr. Tim Frandy gave us the prompt for our essay experiment: What are some of the tensions these days between youth culture and traditional elders in a given cultural community?
We handwrote our reponses to this question, then we immediately ran our writings through various AI / plagiarism detection programs. After writing, Matthew and I debriefed about the experience. Here was the conversation we had, as we thought about what we’d written, and why we’d done so:
David: What makes you curious about this kind of experiment?
Matthew: This issue was first brought forth when a couple of my papers were flagged for AI detection. I ended up doing some research and scanning through some of my old papers before this whole AI craze was a thing and what ended up happening was that some of the papers I’d written before got flagged for the same kind of things. I guess what I’m most curious about is: What about a certain piece of writing causes this kind of AI detection? What constitutes original work? How do we navigate this new territory with so much AI technology around?
David: Do you have any guesses about what’s going to happen when we run these essays we just hand-wrote through a detection program?
Matthew: It’s kind of hard to tell. We did write in a limited amount of time and I didn’t really get a chance to proofread. Usually, I end up erasing half of what I write initially. If we put these essays through, I guess it could be AI-detected; it could be not, depending on how it uses its technology.
David: Do you have any theories about why this happens, or why it may have happened to you? Are you just baffled by it, or do you have any guesses.
Matthew: This is all super new, but from the research I’ve done, AI tends to flag anything written with kind of perfect grammar or phrases that are too good to be written from a human perspective. For me, I do sometimes use a source, Grammarly, which helps clean up your writing. That might have been what caused that AI detection.
With great anticipation, we ran both our handwritten essays through a number of AI detection platforms, including Originality.ai, EasyBib, undetectable.ai, and Grammarly. Undetectable.ai told us that our content appears human, but encouraged us nonetheless to “sign up to humanize and make this text undetectable.” The most interesting and unexpected result was from Grammarly, which told us that “We have found plagiarism in your texts and have also detected 40 writing issues.” Both of our essays were stamped with a bright red “plagiarism found” label. We were told by Grammarly that if we wanted to find out what specific features in our essays were plagiarized, we’d need to sign up for Grammarly Premium.
The takeaway from this experiment was not just about the expensive platforms we’d need to sign up for to check for plagiarism or AI-generated text. It runs deeper than that. I was sad to see the new ways that our own language can become a source of fear, anxiety, and apprehension for us. Not about its merit and clarity or ideas, but about what deficiencies these platforms might find in it, far beyond my control and awareness. Considered together with all the other ways we may experience what Dr. Marie-Eve Bouchard calls linguistic insecurity, it can become harder to use languages in university assignments with a sense of satisfaction, pleasure, and confidence.
But that doesn’t mean we need to give up expressing ourselves in all the ways that come naturally to us. In March 2023, 30,000 signatories from across the professions, including CEOs of many AI firms as well as prominent scientists, joined a call to Pause Large AI Experiments until “we are confident that their effects will be positive and their risks will be manageable.” University teachers and students should take this as a warning that it’s time to pause, take a breath, and figure out who exactly it is we desire to be as writers, dreamers, language-users, and communicators, before we decide we must dive in and start using every new language “tool” that presents itself to us. Our integrity, our principles, our joy, and our learning are always more important than the apparent pace of technological innovation around us.
Related content:
- Thinking about ChatGPT? ChaptGPT and Other Generative AI Tools (UBC Academic Integrity)
- The Opportunities of ChatGPT, co-written by Dr. Laurie McNeill, Associate Head, Undergraduate and Professor of Teaching, Department of English Language & Literatures (UBC Academic Integrity)
- Assessment Design in an Era of Generative AI (CTLT)
- Communicating with students about Generative AI (CTLT)
David Gramling is a professor and head of the Department of Central, Eastern, and Northern European Studies, where we teach German, Polish, Russian, Swedish, Danish, and—soon—Yiddish! David lives and works on the traditional, ancestral, and unceded territories of the xwməθkwəy̓əm (Musqueam), Stó:lō and Səl̓ílwətaʔ/Selilwitulh (Tsleil- Waututh) Nations.
Matthew Chan is a third-year undergraduate student in the Department of Central, Eastern, and Northern European Studies. Matthew lives and studies on the traditional, ancestral, and unceded territories of the xwməθkwəy̓əm (Musqueam), Stó:lō and Səl̓ílwətaʔ/Selilwitulh (Tsleil- Waututh) Nations.
