As a teacher, I don’t mind having to deal with a bit of AI prose in students’ papers. What does bother me are fake AI citations. In this trilogy, I’ll be unpacking them and explaining how to spot them and how I deal with them.
A Trilogy on Academic Slop, Hallucination, and the Death of the Citation.
Rocket-Fuelled Orwellian Flying Pigs: When Reality Out-Hallucinates the AI
Look at the image below.
What’s wrong with this picture?
Can you see what’s wrong here? Do you spot the anomaly?
This is a screenshot of a citation PDF I forced a student to send me before I would release her final grade for an essay. I was withholding the score because I suspected she had over-used AI. And note, I say over-used. Of course she used AI. The question today is simply: how much?
I teach academic writing to English Lit. majors in Japan. Compared to many of my colleagues in similar EFL contexts, I have it lucky. I teach English majors, not engineering students forced to study a language they resent. I teach people who have chosen to be here, who—you would imagine—have an interest in Literature, or at the very least a basic interest in English.
This used to be a lot truer than it is today. It feels “wronger” than ever now.
The aim of my class is simple: research, plan, write, submit. If the essay is 1,500 words and has three citations correctly formatted, it’s a pass. It sounds like an easy feat. Yet, for over a decade, students have struggled with the basic mechanics of a citation. I have tried to teach them. I have told them it’s like a hyperlink (body text mention, full list in works cited). I have pointed them to Google Scholar, and the little “cite” button. But still, many of them fail to master the rudimentary mechanics.
I realized recently it has nothing to do with my failings as a teacher—it is their failure to give a shit. At this point, I have also run out of fucks. Sorry for the expletive, but burnout does what it does, and this is the only way I can truly and emphatically express the emotions of a decade, truncated into a single sentence. So, the new rule is: “get it wrong, F and resubmit.” Since putting the onus on the students to just get it right, life has been marginally more bearable. Since teaching this stuff sent them to sleep, I now just expect them to know it.
But since the aggressive penetration of LLMs into education, this already murky teaching situation has got a lot “siltier.” It’s an icky mess now, and the worst part of that mess is the part where academic integrity truly falls away, due in no small part to the ubiquity of fake AI citations.
Finding Fake AI Citations
The problem is hallucination. AI makes stuff up and calls it a published paper. This means I now have to play detective, checking every citation to ensure the student didn’t just ask an LLM to hallucinate their entire semester’s work. 30×3 citations per class to check and verify… that’s a lot more than my limited F-ratio can take.
So, back to the student. She submitted a paper that smelled like algorithmic slop. She can barely string a sentence together in class, let alone write a robust research paper on George Orwell. When I couldn’t find her source, I assumed it was a hallucination, just another fake AI citations. I demanded the PDF, fully expecting her to fail when she couldn’t produce it. This would prove that she used AI, effectively catching the hallucination in a net.
However, she sent the file. Lo-and-behold, it exists. And that is when I discovered that the problem wasn’t the AI. It was something far worse.
It is the problem of SATURATION.
Look at the screenshot again. This is a paper titled “The Concept of Equality and its Betrayal: Exploring Class Struggles in Animal Farm”. Standard stuff.
Now look at the journal.
Tuijin Jishu / Journal of Propulsion Technology.
Propulsion.
Somehow, in a journal dedicated to rocket engines and aerospace engineering, there is a published article about the proletariat pigs of George Orwell.
WTF.
Why Propulsion? Are Orwell’s pigs flying now?
When are fake AI Citations not fake AI citations?
Why would a journal like this publish a paper like that? I assumed this was a simple matter of fake AI citations, but my student didn’t cite a hallucination, the system made the hallucination flesh, like summoning an actual demon. AI simply helped her find the hallucination faster. But to understand why a student would cite this, and why a journal would publish it, we have to look under the bonnet at the engine. We have to analyse the paper itself.
This is what we call a “Hijacked Journal.”
Sadly this is not just as simple as an AI hallucination or one of those ‘straightforward’ fake AI citations. It’s a zombie. Scammers have cloned the ISSN of a legitimate Chinese engineering journal, built a fake website, and are now charging desperate scholars a fee to publish anything—including literary critiques of pigs—without a single human eye reading it. Learn more at Beall’s List and Retraction Watch.
If the Journal of Propulsion Technology is publishing critiques of Animal Farm, then the authenticity of the entire ecosystem is destroyed. We are worried about the robot taking over the classroom, but we should be more worried that the academic publishing world has already turned into a content farm, fuelled by money, apathy, and—apparently—propulsion technology.
(Continued in Part II: Analysing the anatomy of a Fake AI Citations)
In the next post, I’m going to open up the PDF and perform an autopsy on the paper itself to show you exactly how deep the rot goes.
If you want to see the forensic evidence of the “Flying Pig” paper, make sure you’re subscribed to Uniliterate. I’ll be dissecting the carcass in Part II.
